You're closer to the heart of the issue than anyone else who has written to me about this. I don't question that Andruw Jones was a very, very good defensive outfielder. The issue is compelling evidence. Paul Blair and Garry Maddox were near-legendary defensive outfielders in their day. Is there compelling evidence that Jones was BETTER than Maddox or Blair or Richie Ashburn or Joe DiMaggio?
And the answer is, No. There is no compelling evidence that that is true.
40 years ago and more, Pete Palmer and others were trying to measure the number of runs saved by defensive players, sometimes with good results and sometimes with unfortunate outcomes, but you have to start somewhere. The problem is that fielding stats were so poorly designed, in the 1870s and 1880s, that they don't make any intuitive sense. Henry Chadwick tried to create one set of fielding stats that worked for all nine positions, which is just completely impossible, since the responsibilities of the positions are so wildly different. I have pointed this out before, but between 1880 and 1980 hitting stats added many new categories, and pitching stats added some new categories. Hitting stats in 1880 did not include RBI, or walks by batters, or caught stealing or GIDP or on base percentage or slugging percentage. They did include stolen bases, but the definition of a stolen base in 1880 was fluid and fungible; it was more like the official scorer saying "we'll call that a stolen base, I guess." At some periods a runner might go from first to third on a single, and the official scorer might say "We'll call that a stolen base", or he might not call it a stolen base.
The point is that batting stats, BECAUSE THEY MADE SENSE, could be refined and improved gradually to make MORE sense. They could be, and they were. But the official FIELDING stats, because they made no sense, were precisely the same in 1980 as they had been in 1880. You take an 1881 baseball guide and a 1981 baseball guide and look at the fielding stats, it's exactly the same categories. They had been so badly designed that no one had any idea what to do with them.
Obviously, we are not where we were in 1980. Obviously, we have made tremendous progress in understanding the defensive contribution of each player; not me, but "we" collectively. We as a field. But, having made this progress, having developed multiple ways to measure defensive contributions, people want to say, "OK, we've got it now. We've figured it out. Fielding is now on the same level as hitting, so that we may say WITH GREAT CONFIDENCE that one run saved by a fielder is the same as one run contributed by a hitter." Not talking about the same level of value, but the same level of confidence as to the accuracy of the statement.
But we're not. We're not at the same level of confidence or understanding. There is no way to argue rationally that we ARE at the same level of confidence or understanding. Whereas we might have been at 10% of the same level of confidence in 1980, we might be at 60% or 70% of the same level of confidence now.
Yes, I understand that the people who have studied fielding--John Dewan, and Tom Tippett, and Tom Tango and Pete DeCoursey and Sherri Nichols and David Laurila and others too numerous to mention. I understand that these are very intelligent people, I understand that they have worked very, very hard to measure fielding, I understand that they have been entirely sincere and honest in their work and that they believe absolutely in their conclusions. I don't question any of that, but it is not the same. We grew up with batting statistics and fielding statistics. We relate to them as our native language. Fielding statistics are a language we learned as adults. It isn't the same. It isn't the same in a lot of different ways.
Park effects. We should note, to begin with, that park effects are not as big an issue as they used to be because the differences between parks are not as obvious as they used to be. But do we understand park effects in fielding the same way that we understand them in hitting?
It is obvious that we do not. Do we understand the effects of fielding behind a left-handed or right-handed pitcher. . .do we understand that in fielding statistics to the same extent that we understand it in regard to hitters? Obviously we do not. . .or fly ball pitchers or ground ball pitchers. Do we understand the interactive effects as fully?
When a third baseman plays next to a shortstop who has limited range, he shifts a step to his left. Do we understand the effect that this has on his fielding stats? When center fielder plays behind a great defensive second baseman, he may react a little bit less aggressively to a pop up in shallow center, because he may assume that the Brendan Rogers will get that, whereas Christian Arroyo might not. We just don't really have that depth of sophisitication in our interpretation of the data.
Our experts in this area have created measurements that SHOULD work. They have thought through thorougly the question of "how can we measure this fielder's ability", and, while I do have some disagreements with the decisions they have made, that's not the real issue. The real issue is after-the-fact skepticism.
And we get back to the key concept: compelling evidence. Science requires not merely that intelligent people work diligently and honestly on the problem; it depends on harsh scrutiny from other research. It depends on 50 others researchers trying to tear down the work of the first 10 researchers--and failing. Only after the work has been subject to a generation of harsh scrutiny from people who are sincerely trying to prove that it ISN'T true can the evidence be considered compelling. It's a little bit like a trial. A prosecutor--entirely sincere in his work, entirely honest--may nonetheless become convinced that Steven Avery is guilty of murder, when additional research may show this is clearly not true. It's a prima facie case, a case that appears to be true as long as no one is there to research the other side. And even when the defendant has an attorney to help him disprove the prima facie case, the system still fails sometimes.
We have a prima facie case that Andruw Jones was a greater defensive outfielder than Garry Maddox--and that is ALL that we have. Another generation of researchers may conclude that that was a valid argument, or they may conclude that it was completely wrong. The evidence is not compelling; it is merely designed in such a way that it SEEMS compelling to those who have done the work. I won't live long enough to see the end of that argument, but I am 100% confident that there is another side of the case to be presented.