I stopped reviewing books about fifteen years ago because I was just too mean.   A lot of times, in writing book reviews, I would make both positive and negative comments about the book, but the negative comments would stick, while the positive would wash away.   The author of the book would perceive it as a negative review, not unreasonably or unfairly.   Very often I would feel, months later, that the book was better than I had said that it was.   I just decided that I didn’t need the aggravation, and the authors were entitled to a better reviewer.

                When the first Historical Abstract was published, almost 30 years ago, Lawrence Ritter reviewed it for the New York Times.   Later he would give an interview in which he listed the Historical Abstract among his all-time favorite baseball books, but you wouldn’t have guessed this from the review.   It’s the same syndrome; you may like a book, but there are still things about it that you will not like, and, as a reviewer, you’re inclined to say so.

                There is an ethic that you don’t review your friends’ books.   The New York Times and the other leading book review sources will not ask you to review a book by your buddy; on the other hand, I know everybody who writes baseball books, basically, so neither is it expected that there will be no relationship at all between author and reviewer.   You wouldn’t want to review your friends’ books, as an ethical person, because a) you’d worry about cheating the public, and b) you’d worry about damaging your friendship with an overdose of truth.   It’s just not done—by which I mean, of course, that it is done sometimes, but that it violates the code. 

                We come, then, to The Fielding Bible III, which is a new book by my good friend and business partner John Dewan and his associate Ben Jedlovec; Dewan, by the way, stands for "Defense Evaluated, Weights and Numbers."   Jedlovec stands for "Just Evaluating Defense Lets Offensive Value Escape Criticism", which I take to be a warning against only looking at defensive value.   Anyway, now that you are on notice that John and Ben are friends and that I can’t give you an honest and truly unbiased opinion, I wanted to tell you—and tell them—how much I enjoyed the book.  

                And I can explain why.   One of the standards to which I try to hold myself is that the reader should know something specific, at the end of an article, that he didn’t know before reading the article, and that he should be able to say what it is.   If we suppose that the person who reads the article goes to a baseball game with a friend later in the day, he should be able to explain to the friend what it was that he learned from reading the article—and whatever that was should be interesting enough that he might actually to do so.   Ideally, every paragraph should meet at least the lower standard:  that the reader expects to know something, after reading the paragraph, that he didn’t know before. 

                Of course, that need competes with others; sometimes we have to explain stuff that’s not interesting itself in order to support a conclusion that comes later in the article.   Anyway, the Fielding Bible meets the higher standard a very high percentage of the time.   An article contrasting Derek Jeter with Brendan Ryan demonstrates as clearly as it could be demonstrated why Ryan is better in the field:  he makes the play in the hole.   Jeter and Ryan are almost identical to one another—and almost identical to league norms—in terms of their ability to make the play to their left, to the second base side.    But on balls hit to their right they are as different as Afghanistan and Paris; every step to his right, Ryan’s advantage over Jeter grows larger and larger.   The data is striking.

                An article about Robinson Cano on the double play (sorry to pick on the Yankees). . .an article about Cano on the double play forces us to re-think and re-consider what we thought we knew—an essential goal of sabermetrics.   A couple of articles about pitch location and where the ball is hit take things that we sort of know, intuitively, and put a foundation of solid fact underneath them, enabling these generalizations to be treated as actual fact.   We get, for example, specific percentages for ground balls being hit to the pull side if the pitch is inside, if it is outside, if it is high, if it is low, if it is high and outside, etc.   If a right-handed pitcher throws a right-handed batter a fastball high and outside and the result is a ground ball, the ground ball is pulled 55% of the time.   If the ball is down and in, it is pulled 87% of the time.   There is a chart, reminiscent of the famous Ted Williams chart about what he can hit by pitch location, which shows how often ground balls are pulled by each segment of the plate, and there are many similar charts. This is not merely "fielding" that is being studied here; this is one of the best books ever written about sabermetrics.   

                Don’t overrate or over value, then, this quibble about The Shift.   John Dewan and I have had a running disagreement for five years or more about the effectiveness of using the Ted Williams Shift.    My belief, based on nine years of watching teams shift to try to stop David Ortiz, is that it doesn’t work, and that, while obviously the shift does lead to plays being made on balls that would otherwise get through the right side of the infield, this is offset or more than offset by plays that are lost at other places.   John’s argument, I think, is that if you have to make more plays if you place the fielders where the batter tends to hit the ball.  

                Those are not the complete arguments, and I’ll get back to that.    But The Fielding Bible (III) has a series of articles relating to the shift, The Ted Williams Shift (pages 43-50) and Beyond Ted Williams (pages 51-56).    These articles, again, meet the basic standard: the tell us things that we didn’t already know, specific things that I could (and will) relate to you.   I will treat these articles as if they were a part of the ongoing argument between John and myself on this point, although, of course, that might not be exactly the way a third party would read the articles. 

                John’s approach is to focus on Ground Balls and Short Liners hit by players against whom the shift is often employed.   We learn, for example, that the shift was put in place against David Ortiz 486 times over the last two seasons—or, more carefully stated, that David Ortiz has put 486 balls in play when the opposition was using a shift. 

                Let’s see. . ..Ortiz over the last two seasons has had 1,111 plate appearances, which is (or should be) a lucky number.   He has struck out 228 times (883 left) and walked 160 (723 left).  He has hit 61 homers (662) and has been hit by a pitch 3 times (659).   This suggests that the shift is used against David Ortiz, unless I am missing something, 78% of the time.  

                Sounds a little low, but let’s go with it.  To show the unique importance of David Ortiz in this discussion:  The Shift has been used against Ortiz, over the two years, 486 times, not counting the strikeouts and walks and such like, but the most times that any team has USED the shift over the two years is 437, by Tampa Bay, and no other team is even over 300.    It’s a definition of dominating a category. Remember all those articles pointing out that Babe Ruth hit more home runs than any other TEAM in 1920?  Same thing; the number for an individual is greater than the number for any team.    We should also remember to give The Fielding Bible credit for giving us hard data as to the number of times in a season Tampa Bay uses the shift.

                Anyway, the question I am trying to get to—not the question that John and Ben posed, although they should have—but the question that I am trying to get to is, how many hits per season does David lose by hitting ground balls and line drives into the shift?   John and Ben’s answer, surprisingly, is "none".    Their data shows that David Ortiz’ batting average on ground balls and short liners over the two years has been .245 when the shift was used, and .232 when it was not used.    Their explanation for this is "Well, teams shouldn’t be shifting against David Ortiz anyway."  We only recommend using a shift against a hitter who pulls his ground balls and short liners 80% of the time.   David’s not at 80%, so we wouldn’t recommend using the shift against him. 

                It would be difficult to document all of the reasons that this strikes me as an inappropriate exclusion, but let me try:

                1)  If you concede that the shift doesn’t work against David Ortiz, aren’t you conceding that the shift doesn’t work?

                2)  Although we have no actual data, I am certain that the shift has been used against David Ortiz far more times than it has been used against any other hitter in the history of baseball, including Ted Williams (since it didn’t become common against Williams until the second half of his career.)   Suppose that I were studying the effectiveness of the stolen base, but I found some way to exclude Rickey Henderson from the discussion, or suppose that I were evaluating the intentional walk, but I found some way to say, "Oh, Barry Bonds doesn’t count; you shouldn’t be walking Barry Bonds anyway."   Wouldn’t you say that this damages the credibility of the analysis, to exclude Rickey Henderson from a discussion of the stolen base, or to exclude Barry Bonds from an evaluation of the intentional walk? 

                3)  If you are evaluating whether the shift works, you have to evaluate whether the shift works.   You can’t gerrymander the data by carving out some segment of the data in which it appears that the shift works.

                4)  Even if it were a legitimate approach to carve out a subsection of the data (which it could be in some cases), and even if there were not special problems with excluding David Ortiz from a discussion of the shift, I simply don’t believe that there is a legitimate basis to exclude him.   What’s the actual data which is the basis of this exclusion? 

                I’ve seen David Ortiz play almost every game, live or on TV, since 2003.   He pulls the ball quite a lot.   Yes, he did go to the opposite field more in 2011 than he had in the previous 4-5 years, but there was no way the opposition could have known this would happen going into the 2011 season.   They’re essentially saying that the opposition should not have shifted against David Ortiz in 2011 because the 2011 data shows that he doesn’t pull his ground balls 80% of the time.  

But present strategies are always and necessarily based on past data.   What they’re doing is a step toward saying that the opposition should have shifted against David Ortiz when he was going to pull the ball, but should not have shifted against him when he was not going to pull the ball.  

                We are drifting here into one of my auxiliary arguments against the David Ortiz shift:  it’s irrelevant.    Look at David Ortiz in 2006:  54 homers, 29 doubles (probably 15 of them off the wall, let’s say), probably 20 others singles off the wall, 119 walks, 387 balls in play.   What is it you think you’re going to accomplish by moving the fielders?   His expected average on balls in play is probably .300; let’s say you move it to .270.   So what?   54 homers, 119 walks, some other balls hit off the wall. …those are the things that make David Ortiz deadly.   The Shift only makes a difference if the pitcher gets the result he wants.   

                That’s an argument that the shift is not very effective against Ortiz because NO defense is going to be very effective against him; it’s not an argument that the shift is not more effective than another defense.   What we’re arguing about here is the minor point, whether a shift is more effective than another defense.   Getting back to that argument. . ..

                OK, here’s what we really need to know.   We need to know what the results were against David Ortiz (and other shifted-against hitters) when the shift was used, and when the shift was not used.    If you just studied that, you would find that there was no meaningful difference.   John (from my standpoint) is attempting to show that there is a meaningful difference by drilling down on a subsection of the data, even though, to accomplish this drilling-down effect and find what appears that it might be a meaningful difference, he has to exclude David Ortiz from the study, which is a bit like excluding voters from a study of elections.   But the process of drilling down to the most relevant data, setting aside that problem, is a legitimate and constructive process, and I’m not arguing about that.

                John wants to focus on ground balls and short line drives, which, again, is a legitimate and constructive step toward understanding the problem, even though I think it is being used to create an exaggerated estimate of the shift’s effects.  The first point we need to reach is that of knowing how many hits Ortiz loses by hitting the ball into the shift.    The article deprives us of any answer to that question by excluding Ortiz, and also by not actually reporting the most relevant data. 

                1)  How many ground balls and short liners did Ortiz hit against the shift?

                2)  What was his batting average on those?

                3)  How many ground balls and short liners did Ortiz hit against "normal" defenses?

                4)  What was his batting average on those?

                If we had that data, for all of Ortiz’ career, we would really have something.    We don’t have that data, because

                a) nobody was keeping track of when the shift was in use until the last couple of years,

                b) nobody was keeping track of ground balls and short liners at the start of Ortiz’ career, and

                c) John and Ben didn’t choose to present the relevant data that they have and should have presented to us for our evaluation.  

                We are left, then, to speculate on what this data might be, based on the facts that we do have.   How many balls does David Ortiz put in play, in a season?  About 380; 380 is the average of the last two seasons for him.   Most of his data in those two seasons is about the same, the difference being that he struck out less in 2011, thus picked up some additional hits. 

                380 balls in play?   How many of those are ground balls and short liners?

                About 190, we can guess.   Some of their charts (which accompany their article) inform us that 46% of Ryan Howard’s last 250 balls in play were grounders and short liners, 50% of Prince Fielder’s, 51% of Aubrey Huff’s, and 52% of Uggly Dan’s.   50% seems like a norm, so let’s assume that David Ortiz hits 190 ground balls and short liners in a typical season.

                How many hits does he lose by hitting the ball into shift?    Here John and Ben’s article suggests an array of different percentages—30 points in batting average (on those plays), 51 points, 38 points, 43 points, and 61 points.    The 30-point figure is calculated including players who beat the shift with a bunt, and I can see legitimate reason to exclude those hitters. . .

                Or not.   I could make the argument here that if you’re going to study the effects of the shift, you can’t study the effects of the shift except when it doesn’t work.   I could make that argument, but we’ll skip it.   The 61 points seems pretty clearly illegitimate, since that is derived by excluding from the study David Ortiz and everybody else who messes up their good data, giving them only a handful of players left IN the study—5 players, who, over the two years, have put a total of 1,266 balls in play against the shift.  

                You can’t take that number seriously, so let’s say 50 points, which is being very generous, but let’s say 50 points.   Nine or ten hits.   David Ortiz loses nine or ten hits a season by hitting ground balls or short liners into the shift. 

                As negative as I am about this study, I will concede that having this much information—having the information necessary to make this estimate—is a contribution to the discussion.   We understand the problem a little bit better now, assuming that this estimate is reasonable, than we did before the study was published.  

                Now we come to the real problem with the study.   So far I have just been pit nicking.  

                Let me ask you a question:   If you were trying to evaluate a strategy, would you:

                a)  try to measure the positives and the negatives, so as to compare the two, or

                b)  only try to measure the positives, and totally ignore the negatives?

                Option (b), apparently; that’s the model that John and Ben have chosen for this article—to totally and absolutely ignore the negative consequences of using the shift, while focusing all of their attention on measuring the benefits.  

                At the conclusion of their article, Ben and Johnny (sorry) suggest eight questions about the shift which require further study.   In this list, they don’t hit any of the real questions that, in my opinion, actually need to be studied—suggesting, to me, that they just really do not have any genuine understanding of the problem.    They want to ask stuff like "Should pitchers adjust their pitch selection and location with a shift on?"   You mean like, throw David Ortiz inside fastballs because you have a shift on?    Oh, right; please, PLEASE throw David Ortiz more inside fastballs.   Let’s do everything we can to encourage that.    I assure you guys:  The Shift will protect you; just throw him low inside fastballs, and you’ll be fine.    

You don’t alter your pitching pattern against Jim Thome to control where he hits his ground balls, because you’re not concerned about his ground balls.   You’re concerned about the 400-footers.   You can alter your pitching pattern to move the ground balls for Fred Lewis and Felix Pie. 

                The real problems with playing The Shift are

                1)  That it leaves sections of the field undefended, and

                2)  That players are playing out of position, which very often requires them to do things that they don’t exactly know how to do.

                All kinds of weird things happen, because of The Shift, that would never happen if you were playing a normal defense.   The variety of these would make them difficult to measure.  A lot of them are things that only happen once, but. . .they happen.   Third baseman playing over near second base; ball grounded into the hole, the third baseman (playing short) dives for it and knocks it into foul territory, runner scores from first and Ortiz reaches second on what is basically a GB6, because the third baseman doesn’t know how to make the play and the person who should be in position to contain the damage isn’t where he normally would be.     

                Here’s a play that happened three times in 2006-2007 (could be wrong about the year.)   Pedroia (twice) and somebody else (one time) stole second, with the shift on, then popped up and headed for third base because nobody was guarding third base.     That won’t happen against Tampa Bay, because Tampa Bay plays the shift all the time, so the players understand that somebody has to cover third base on a steal of second.    But a team that doesn’t play The Shift a lot—it will happen. 

                These are among the questions that I would like to see answered, before I am convinced that The Shift is, on balance, a positive for the defense:

                1)  How many outs do you lose, playing the shift, because foul pop ups drop uncaught wide of third base?   This is not something that happens once or twice a year; this is something that happens on a pretty regular basis.    If a left-handed power hitter gets hit on the knuckles, one of the most frequent outcomes is a pop up outside of third base.   If there’s no third baseman, there’s no play.  

                2)  How often does a bloop or flair—a soft pop up—drop behind third base because there is no third baseman?  (Sometimes even a high pop up will drop behind third because the third baseman is so far away that he can’t get there in time to set up underneath the play.)

                3)  How many double plays do you lose because the second baseman sets up so deep that you lose two steps before he fields the ball?

                4)  How many double plays do you lose because the third baseman finds himself the pivot man on what should be a 4-6-3 double play (which has become a 6-5-3 double play because everybody is out of position)?   (Also, of course, a 1-6-3 double play becomes a 1-5-3 double play, which creates the same problems, and the 3-6-3 double play has become a 3-5-3 double play.)

                5)  How many Defensive Misplays and Errors occur because the third baseman is out of position, and playing a position he doesn’t really know how to play?

                6)  How many Defensive Misplays and Errors occur because the shortstop is out of position and playing an unfamiliar position?

                7)  How often does the team screw up a relay play because players are out of position on the relays?

                8)  How many weird, one-of-a-kind plays are there (working to the advantage of the offense) because the defense is misaligned? 

                It seems like there are a lot, but I haven’t really made a list of them.    I doubt that we can measure that (number 8) because it’s so hard to define "a weird, one-of-a-kind play".    I’m not sure we can measure (5) or (6) because the number of plays on which The Shift is used is probably not large enough to measure reliably changes in the frequency of defensive misplays per position.  

                I don’t believe in The Shift, and, as I said, I don’t believe that John and Ben have any real understanding of the problems that The Shift causes for the defense.   But I don’t mean to overstate the importance of this issue relative to The Fielding Bible as a whole.   The Fielding Bible is a tremendous book; I am merely arguing with this one article, or two—and even in that context, John and Ben have made some contribution toward a better understanding of the problem.