Aurametrics: Can Musicians be Quantified?

I love baseball. There's so much to love. It combines the athletic abilities of running, throwing, catching, and intricate applications of hand-eye coordination with a cerebral, strategic mind-game that fluctuates from pitch to pitch. Baseball is a sport where excellence is not limited to the physically gifted; there are no height requirements like in basketball, or weight/strength requirements like in football, but as Hawk Harrelson consistently reminds us, "it takes all shapes and sizes to play this game." The uniqueness of baseball doesn't end there, as Ken Burns likes to point out in his documentary and in interviews, baseball is one of the only sports in which the defense controls the ball, and the majority of the action. It is also exclusive to baseball that the players, not the ball, accumulate points (or runs). Not to mention that, like any sport, baseball has a rigid set of rules, however, unlike other sports the actual dimensions of the playing field vary based on the city or stadium in which the game is being played.

None of these fascinating characteristics accurately describes why baseball is my favorite sport though. I will continually be riveted by baseball because it is the ultimate team game, yet it is made up of a series of individual encounters, or "battles" as Jeremy from Sports Night would say. Every play starts with a "battle" between one pitcher and one batter, and what happens next is almost completely determined by the skill of those two individuals. Despite the individuality of this encounter, baseball truly is the ultimate team sport because a batter is only allowed to hit once through a lineup, which results in about 4 or 5 at-bats per game. It's not like basketball where you can go to your best player on the most important crunch time plays, or like football where you can feed the ball to your star RB or throw a fade to your monster WR. I'm sure Tony LaRussa would love to send Albert Pujols up to the plate in every important scenario, hell, I'm sure he'd like Albert to take every at-bat, but baseball doesn't work that way. However, since we have so many individual encounters, statisticians are able to quantify the sport more than any other.

With the technology used these days, we are able to track the usefulness (or lack thereof) of literally every action on a baseball field. Sure, advancements are still being made to analyze fielding (though the UZR and DRS systems are a pretty solid indicator with a large enough sample size), or pitching, but anything a hitter does can be tracked, counted and analyzed to a point that doesn't exist in any other sport. Surf around the Fangraphs glossary sometime, especially the offensive statistics or miscellaneous statistics and you'll find a maniacal thoroughness in their newly invented statistics: http://www.fangraphs.com/library/ Some might find this sort of analysis boring, but personally I love numbers, and I love being able to figure out why or how things happen. This sort of analysis is like brain food for me, it's enthralling and satisfying all at once, and it supplements my love of a game that I already appreciated anyways.

So what does this have to do with the title of this blog? Well, as I mentioned in my inaugural blog post, I'm a professional piano player, and I was playing for a wedding this past weekend. I love playing weddings or parties, or any kind of celebration because everyone is always in a good mood and happy to be there, not to mention that there's always free food and usually an open bar. I'm completely accustomed to people coming up to me after I play at the event and giving general compliments like "great job!" or "the music was beautiful!" to the point that I'm kind of numb to it. I mean, it is my profession, and I take my playing seriously so I practice and prepare so that I will sound good, the same way that I'm sure these people are thorough with any job that they're being compensated for. The difference is that as the song "There's No Business Like Show Business" states:

"The butcher, the baker, the grocer, the clerk

Are secretly unhappy men because

The butcher, the baker, the grocer, the clerk

Get paid for what they do but no applause."

There was one older gentleman however who came up to me and said "I heard you hit a wrong note in there. But you did a beautiful job, everything else was real smooth."

Don't get me wrong, I'm not upset with this man because I always appreciate honesty, and I'm admittedly bored with the same compliments, not to mention I know that I play several wrong notes any time I sit down at the piano and play for a while, almost all of which I assume are unnoticeable to the untrained ear. But his statement got me thinking, Which note was he thinking of? How many notes did I miss, was this the only one? When I perform, what percentage of notes do I miss? What percentage of my mistakes are noticeable to an untrained musician? and lastly, and probably most importantly, Would this ever be worth quantifying?

I realize that this is the exact line of thinking that Bill James had at some point when he pioneered the sabermetrics movement in baseball. I'm sure these kind of thoughts popped into Voros McCracken's head when he developed his DIPS theory. http://en.wikipedia.org/wiki/Defense_independent_pitching_statistics The problem is, I don't believe the technology exists (nor do I feel it necessary to devote the time) to gather the information needed to calculate these things, unless someone wants to develop whatever electronic sensors Bugs Bunny used to have when teaching piano lessons (and the student would get electrocuted if he played a wrong note). But, if the technology existed to make aurametrics (a word I made up today meaning the analysis of music through objective, empirical evidence) possible, here are the stats (that I also just invented) I'd be intrigued to analyze:

Note Percentage (NP): This would be very similar to fielding percentage (FP), it's just the percent of notes that are hit correctly. In baseball, a solid FP is around .985 to .995 (or 98.5% to 99.5%), and I'd like to think that professional musicians are in that ballpark (pardon the pun), so out of every thousand notes I'm playing, I'm missing 5-15 of them. Sounds about right, on a good day at least. But similar to FP in baseball, NP could be a misleading stat. FP is deceiving because errors can be misleading since they don't factor in things like defensive range, and the importance of the errors isn't weighted. The latter is the biggest issue with NP, the importance of the missed note isn't weighted. Some missed notes are greater offenses because they're more obvious to an untrained ear than a minor mistake that most wouldn't notice, for example the note that I missed that provoked the old man at the wedding to mention it to me. This would prompt aurametricians to come up with the following stat:

Discernible Note Percentage (DNP): Different than a DNP in sports (did not play), this stat works the same way NP does, except it doesn't measure the actual missed notes, but the number of missed notes that the listener notices. It's more useful than straight NP since the only mistakes that matter are the ones that actually affect the listening experience, but this varies from listener to listener. This variance presents aurametricians with the issue of coming up with a consistent baseline of listeners, a constant test group to compare data against, which leads to the next concept:

Replacement Listener (RL): Based on the idea of a replacement player in baseball, which is a theoretical player who can be signed to the league minimum contract and will perform at that level (i.e. about as good as the best minor league AAA players would perform in the majors). Here's more info about the replacement player compliments of fangraphs: http://www.fangraphs.com/library/index.php/misc/war/replacement-level/ A replacement listener would have the average musical acumen of someone with no musical training, and would have to be calculated through an absurd number of aural tests for average people who have had no musical training. This baseline would lead us to statistics like this:

Discernible Replacement Listener Note Percentage (DRLNP): If I'm playing a Chopin waltz, and my fiance is present, she will be fully aware if I make any mistakes, because she's a professional pianist as well, and much more proficient with classical music than I am, so her discernible note percentage for me will be much lower than probably anyone else in attendance, and therefore not a great indicator of the importance of an error in relation to the average listener. That's why DRLNP would be so important, because some mistakes are inconsequential (like an error where there runner doesn't even end up scoring in a 12 run blowout), but some mistakes are obvious and important. This statistic would accurately reflect the mistakes that I make that would be obvious to anyone even if they don't have any musical training at all (our RLs).

If we're taking into account the importance of a given missed note, it's probably also important to take into account the importance of the situation of the performance too. Like comparing a baseball player's spring training stats to their regular season stats, or their regular season stats to their postseason stats. So here's a few more NP-based statistics:

Practice Note Percentage (PNP): The NP of notes played while practicing, or more specifically, not in front of a live audience.

Live Note Percentage (LNP): The NP of notes played in a performance in front of a live audience.

Of course we can apply these statistics to our replacement listener as we did earlier, giving us the following stats:

Discernible Replacement Listener Practice Note Percentage (DRL-PNP): Just calculating the DRLNP in practice situations.

Discernible Replacement Listener Live Note Percentage (DRL-LNP): Calculating the DRLNP in a live, performance situation.

Using these stats we can see how well a musician can deal with nerves, or how well they perform in a live performance situation. Translated into sports, these stats give us a good feel for how "clutch" a musician is. How else could we determine who the Derek Jeter of music is? World-renowned jazz guitarist Bobby Broom once told me (and I'm paraphrasing) that a musician will probably play to about 80% of his ability in front of a live audience compared to the 100% of his ability that this musician plays at in a practice room, mostly due to nerves. He had no statistics to back this up obviously, because the point was that any musician (or athlete for that matter) needs to over-prepare to be ready to deal with the nerves of a live situation, a circumstance that is difficult to simulate. If aurametricians are able to compile the data needed someday, we'll be able to actually calculate the effects of nerves more precisely, but with this information we could still calculate a pretty fascinating ratio to see which musicians deal with their nerves the best. Here's my final (and most interesting) statistic of the post:

Clutch Note Percentage Ratio (CNPR): Calculated by dividing Discernible Replacement Listener Live Note Percentage by Discernible Replacement Listener Practice Note Percentage (DRL-LNP/DRL-PNP): The hyphens are not subtraction signs, they're just there to make the six-letter stat names easier to read. Obviously the actual names or abbreviations of the stats might need some work since they have a lot of letters and are lacking the convenient abbreviation with a cool ring to it, like VORP or WAR (again, check the fangraphs glossary if you wanna know what VORP or WAR are). For the purpose of simplicity, I'm going to call this stat Clutch Note Percentage Ratio (CNPR) until I come up with a better name for the stat, or until a better one is suggested. 

The name is not strong, but I believe the premise is. This stat separates the important mistakes from the inconsequential ones through the application of our Replacement Listener, and allows us to analyze which musicians are able to perform best when it counts (in a live setting). A CNPR of 1 means the musician performs with the exact same acuity whether practicing or performing. The higher the CNRP is, the more "clutch" (or accurate in live settings) a musician is when it counts. If the number is lower, it means the musician is unable to play up to his/her abilities in live situations, which can be most likely attributed to nerves. I would imagine CNPR of higher than 1 would be rare, as most musicians don't play better when the spotlight is on them, but it would be fascinating to find out. Now, if anyone has any ideas on how to compile the data, I'm all ears...