My Very Subjective Problem with WAR

A long time ago, relatively speaking, a really smart guy named Aristotle proposed, among many other thoughts and concepts, three modes of persuasion.  His modes of persuasion were ethos, pathos, and logos.  Ethos is an argument made effective given the reputation or authoritative nature of the speaker.  Pathos derives its persuasive value from its ability to appeal to a listener’s emotions.  Logos, the third mode, applies logic and reason to the particular situation under consideration.

In baseball circles, logos is most often employed as a persuasive tool.  Very likely, readers are better because of it.  Yes, humans are emotionally charged and additionally respond to authority, but baseball offers a refreshing lens through which information may more often than usual be digested logically.  Despite these considerations, this post is largely devoid of logos, and instead rooted in pathos.

This post was borne from two occurrences.  First, my coming across an insightful, although seemingly innocuous, quote in “The Undoing Project”, by Michael Lewis.  Second, a conversation I had wherein the etymology of baseball’s WAR statistic was considered. 

The quote in question is this: “In many ways I was a glorified secretary, and that troubled me for many years… Deep down, I thought I was extremely replaceable.”  

The man responsible for this quote is named Don Redelmeier; without giving anything away, here he expresses his insecurity working alongside a particularly brilliant man.  Redelmeier is an MD, he has a strong CV, a significant list of publications, and is interesting enough to be a subject in a Michael Lewis book.  For all intents and purposes, he is quite successful.  Yet despite his abilities and successes, the possibility of being replaceable was something that gnawed at him for not days or weeks, but years.

In that conversation regarding WAR, the individual who I was speaking with mentioned that, in his opinion, it is a fairly perverse statistic.  Perverse, not because of a gross accounting error in talent evaluation, but because players are quantified by a now-common catchall statistic which frames them each in finite terms of how replaceable they are.  I was caught off guard by this point made by another fan.  His complaint was not one of logos or even ethos, but pathos. 

“Deep down, I thought I was extremely replaceable.”  Wins Above Replacement.

WAR is, and has been for a number of years now, a ubiquitous statistic.  There are several mainstream iterations of WAR in baseball; it is the most common measuring stick of player quality when a quick number is required; its concept is quickly finding mainstream traction in other major sports.  WAR is so common and widespread that I simply had never deeply considered its verbiage, only its numerics.

The first purpose of this post is to very briefly and incompletely compile some resources chronicling the inception and evolution of WAR. The other purpose is to play devil’s advocate to the standard jargon, and suggest that upon deeper consideration, “replacement” is a particularly crude word to be included in WAR’s composition.  

Tracing the origins of WAR is at least a little bit tricky, in part because its creation did not occur overnight.  To use a poor analogy, WAR, like Rome, was not built in a day.  Part of the reason for that time intensive process is the fundamental irony of WAR: it can be presented as an easily comprehended, normalized, standalone number only because it has been so thoroughly adjusted and refined on the backend.  

As it turns out, the skeletal structure of WAR began with the exact bone I have to pick with it: the “R”.  According to this article from Baseball Prospectus, WAR as it is known today evolved first from the nascent idea of a “replacement” level player.  The usual suspect, Bill James, produced his abstracts throughout the 1980’s wherein the definition of replacement level became increasingly lucid.  As the aforementioned articles author Brandon Heipp puts it: “whether the concept of replacement level originated with James is something I cannot answer, but it can be traced back to James at the very latest.” 

In addition to James, other key players contributed both individually, collaboratively, and sequentially to make new insights and breakthroughs.  Tom Tango, who has himself worked directly with WAR, points out here how Pete Palmer (WAA) and Clay Davenport (WARP) made breakthroughs of their own which aided their peers and popularized this train of thought.  VORP is an example of another tool used for placing players in the replacement level context.  James’ abstracts illustrate a multi-year, multi-stage process.  Overtime his Runs above Replacement (RAR) metric translated to Wins above Replacement for an even more easily digestible catchall.

These various iterations previously cited increasingly incorporated nuanced facets of the game such as base running, park factors, league adjustments, and positional adjustments as well.  Today, there appear to be four primary models to define baseball war; they originate from FanGraphs, Baseball Prospectus, Baseball-Reference, and openWAR.  These methodologies each have their own twist and surely various originations.  For instance, Baseball-Reference credits its WAR stat as being “Developed by Sean Smith of BaseballProjection.com.”  

As stated previously, my perspective on this point is one of pathos, not logos.  My issue with WAR lies in the semantics.  Semantics may not seem to be of great importance, especially in a world that prioritizes logos, but they are.  They are because words frame how we perceive the information we collect; they are because words can have an ostracizing effect.  

Baseball is incredibly difficult.  The acronym WAR fundamentally undermines that reality.  Baseball Reference writes that a replacement level is “a player that would typically be available to replace [an injured regular]”.  If the odds of a high school baseball player eventually ending up on a 40-man roster are in fact 1:764 (0.0013), it should be made clear that those players are anything but typical, they are already safely entrenched somewhere on the far right side of the bell curve.  I have pointed out before how difficult it is to even break even in terms of WAR even for those first round draftees projected to have the brightest futures.

So using a term like replacement creates a framing issue.  Replacement, as a term, frames ~0 WAR players as not being proficient ballplayers, despite the fact that they are statistically significantly proficient at baseball.  Additionally, replacement infers that there are other suitable players readily available to do the replacing.  In fact, there might not be.  83 MLB players recorded either 50 IP or 170 PA’s (roughly a third of a season, by “qualifying” standards) in 2018 according to FanGraphs while registering negative WAR.  If replacement level players were typical, as Baseball References prior definition suggests, there would not be nearly three sub-0 WAR players per MLB team last year.

While it may be of sabermetric use to evaluate players based on that lower threshold of MLB quality (opposed to average MLB quality, for instance), players are in no need of another explicit reminder of their fragile circumstances.  In a profession defined by the attrition of its athletes, highlighting the precarious nature of their roles could understandably push players to spurn sabermetric insights. 

Characterizing players based on their replaceability broadens the dichotomy between players and new age front office personnel; it creates tension between the old guard and the new within clubhouses everywhere.  A player already wary of the statistics created to quantify their production cannot begin to embrace sabermetrics when the language used to express their talents in fact undermines their accomplishments.  These players dreamt of hitting home runs and getting the final out, not accumulating WAR only to distance themselves from replaceability.

Very simply, baseball is great in a large part because it can be quantified and interpreted through logos-oriented modes.  But it is the players who produce and breathe life into those numbers; I personally dislike the idea of so explicitly framing them in terms of their replaceability.  I would simply prefer a new term, whether it be Wins Above Baseline or Wins Above Proficiency, because we shouldn’t be so wedded to this acronym that it may very delicately undermine the players we all so enjoy following.