Or "The Depeche Mode Strategy for Running a Hockey Team" (Get the Balance Right)
I'm just going to come out and make this outlandish-sounding statement: having managed a large team of highly skilled technical professionals for several years now, I think I have some insight into why objectivity and subjectivity need to coexist in the day-to-day operations of a hockey team. The tricky part is achieving the right balance.
In my company, my position is something like head coach, and my boss is something like a GM. Thanks to JenLC's Twitter summaries of Stan Bowman's participation in the Hockey Analytics panel of MIT's 2011 Sloan Sports Analytics Conference, I noticed some striking similarities between how I run my team and how the Blackhawks are run. Stick with me here.
Using imperfect analogies to hockey, I was a position player for many years; I made the big league fairly young and did a lot of learning on the job. I was never among my team's superstars, but what I lacked in innate skill I made up for with smarts and "soft skills." Eventually, as happens to technical professionals from time to time, I became a manager. I'm not the most technically knowledgeable person in the organization, but I know where and how to get reliable information and have a pretty good bullshit detector. My employees are highly skilled specialists, generally very driven, and tend to have large -- sometimes fragile -- egos.
In general, my boss is the "big picture guy" and I handle the day to day stuff. We have the same objectives, but very different perspectives. Fortunately, it usually works out for the best because we understand the importance of each other's perspectives.
The coach's perspective
The coach has to field the best team for the task at hand, given the current opponent and its strengths and weaknesses. What makes this complicated is that he can't just match player to opposing player or strength to opponent weakness. That would be easy and probably even fairly easy to automate. He has to assemble a team that works as a cohesive unit out of pieces that don't necessarily fit together nicely like a jigsaw puzzle. He's got massive amounts of data at his disposal, both objective (video, stats, etc.) and subjective (what he sees during practice and games, what he knows about personalities and styles, what he hears directly from his players and indirectly from their chit-chat, etc.). Obviously, not all of that data is equally valuable; organizing and taming it can be a huge task itself.
The coach is working toward multiple goals simultaneously, too. Of course, The Big One is a championship, and all others are really just "sub-goals" of that one: finish with a high seeding, make the playoffs, have a good month, have a good week, have a good next game, have a good next practice, keep motivation high but also keep morale high, keep guys happy but focused, keep confidence high but egos in control. It's an enormous job and sometimes coaches have heart attacks and strokes.
The GM's perspective
The GM's ultimate goal is the same as the coach's because it's the goal of the organization as a whole. It probably varies quite a bit around the league how much say he has in what goes on day to day. But the GM (probably) has the most control over what puzzle pieces the coach has to work with. He might also have a lot of input into where those pieces belong. He wants noticeable, measurable results, and he is focused on optimization. He (again, probably) is not too bothered if the team has a bad game or even a bad week or more, as long as the "big picture" trend is still in the right direction. He looks through a much wider angle lens than the coach, so small fluctuations should bother him less.
He is much more likely to be guided by objective data, because he doesn't necessarily have access to the amount of subjective data that the coach has -- and because the subjective stuff is less relevant at the "big picture" level.
The balancing act
Even though I'm a strong advocate for stats-based analysis, my premise here is that a team needs both the objective and subjective viewpoints, in the right proportions (which may be different for each coach/GM combination), in order to be effective. And, to get there, they need a smart coach and a smart GM who trust each other. This doesn't mean one never questions the other's decisions, or that either has 100% freedom to act. Ideally, the coach has a great deal of leeway and appreciates the input he gets from the GM. The GM is confident that the decisions he makes will usually be well received by the coach, because he gets useful, frequent feedback. Both are free to disagree with the other, because they recognize that their goals are the same and constructive disagreements can lead to deep discussions which can lead to novel solutions.
One of the things I identified with most out of the MIT panel was the statement (paraphrased) that the coach values "coachability," predictability, and consistency. As a manager, I wholeheartedly agree. My priority is to build and maintain a high functioning team that, at least from the outside, appears to work as a single unit. The less babysitting of individuals I have to do, and the fewer surprises I get, the easier my job is. Conversely, the more time I have to spend managing individuals and personnel issues, the more likely it is the team as a whole will suffer. I can completely sympathize with the coach who loves smart, dependable "role" guys, "good in the room" guys (of course, in my case, they're not all guys). They make the job so much easier. And, until I was a manger myself, I couldn't imagine just HOW much easier.
According to Bowman, the job of the analytics staff is to balance that subjective view with objective data. A smart coach will recognize that and incorporate that data into his decision-making processes, because he can see that it fills holes in his own assessments.
The tricky part is striking the right balance. A coach's job would be very different if he could coach solely according to objective analyses. Team would always be able to ice their ideal combinations; their only limitations would be talent levels and health. But people are messy and unpredictable. A manager's day is full of compromises, and I imagine it is often the same for a coach. When I have a project to complete, I'll probably think of a "technical dream team" who, based on skills alone, could knock it out of the park in no time. Sometimes, though, I can't put my #1 Component X guy and #1 App girl on the same project. Maybe a more important project needs one of them; maybe they communicate so differently that it slows the whole project down; or maybe they just -- for any number of reasons -- don't work effectively together, and it would be an unreasonble drain on my time (I have five or six other important projects in the works, after all!) to give that part of it the attention it would require.
I'm not saying I think situations like that on an NHL team are common -- particularly on a successful team, where it must be easier to be generally happy and not let little things bother you. But I bet it happens regularly on a small scale (minor playing style or personality differences, for instance), and it's the coach's job to fix or work around them. Sometimes that means he has to put together the best team possible when some ideal-looking combinations aren't available to him. If that's good enough, the GM will still be happy with the results. If not, maybe the coach and GM talk about it and try out some different things.
San Dimas High School Football Rules!
Or: "I couldn't think of a good heading for my conclusion."
If you're still reading, wow! I realize that this post is heavily skewed toward the coach/manager/"subjective" view, but that's the realm I work in and know best. It is definitely not my intention to argue against the utility of objective (stats-based) analysis, because I think they're wonderful -- and fascinating -- tools, and the smart teams are the ones ahead of the adoption curve.
My point, which I've taken the long way around reaching, is that, as long as it's human beings (yes, even Jonathan Toews) we're analyzing and evaluating, some subjectivity will alway be necessary. Smart teams will get the balance right, though, driving by the numbers where they can, and recognizing the spots where course correction may take the form of a little "people management."
I have a bunch of notes on related topics that I could write about or just chat about if you're interested. I'd love to hear any feedback you have, either in comments here or on Twitter (where I'm also The_FFF).