It seems to me this is a continuation of his first post and sets up why he loves looking at stats and how they are useful.

‘cause, yes, anyone could see that. But people would pooh-pooh that with the notion that the bad defense didn’t really matter much. UZR and other metrics show that the shortcoming at defense do matter enough to make the offensive contribution not compelling.

I think one thing people had forgotten, in terms of signing Dunn — the Cubs had potentially two or three good defensive replacements for Dunn (Johnson, Fuld, and early on, Gathright). You’d presumably take Dunn out after his final AB in the 7th inning (home) or 8th inning (road) and have good defense for at least two of the nine innings, thus reducing any damage he could do defensively.

Shawn will get to this soon – the issue of how to properly valuate a player is the hardest (and oldest) problem in sabermetrics. He was just making a more general post here. Suffice it to say though that Dunn’s defense is so awful that it DOES cancel out a good portion of his offensive production

i would say that you cannot calculate the number of runs you would prevent and outs you can make with a better defender that has more range, can throw, and can be quicker to the ball than Dunn. He may drive in 100 runs for you but he will also allow more than a good defender would because of errors, and other bad fielding practices.

Okay, statistically, how good or bad is a healthy Soriano defensively? Throw the numbers out from 2009 because he was hurt so much. I know this has been a source of great debate on this site since Sori got to the Cubs. How much value is there in an outfield assist, which would work in Sori’s favor? Is Sori’s perceived lack of ability to get jumps on balls in the outfield real or imagined? What is Sori’s statistical range? According to my admittedly very fallible, very unscientific, eye test, it doesn’t seem Sori’s range is that great. Then, again, left fielders don’t tend to have that much range in the first place. If they did have a lot of range, then they wouldn’t be playing left field.

The consensus seems to be that Dunn would play RF, leaving Soriano in LF. But if you put either player in RF, it magnifies their defensive shortcomings as that position is more difficult. For example, if the LF misplays a ball down the line, that becomes a double but a similar mistake by a RF leads to a triple, at least some of the time. Perhaps that example oversimplifies the problem, but the fact is that your taking a guy whose defense is costing his team runs in LF and moving him to a more difficult position, essentially exacerbating the problem.

This also raises another question, which might be better directed at Shawn, of what effect the CF has on the defense of the corner outfielders. Call it the Maddox/Luzinski problem … does having an elite CF with outstanding range, such as Garry Maddox, have much effect on an atrocious LF like Greg Luzinski, who was actually a very similar player to Adam Dunn? Certainly it doesn’t matter on balls down the line or over the LF’s head, but what about balls that are hit into the gap? Having a great CF to cover-up some of the LF’s mistakes would matter at least occasionally, but would those instances be frequent enough to show up in the LF’s statistics? While both Kosuke and Reed Johnson are competent but not outstanding in center, would the value of playing a better defensive player, such as Sam Fuld or Mike Cameron, be magnified even further because he’s helping to cover for a crappy LF?

In terms of their value the past three seasons, keep in mind that Soriano’s total of 7.8 WAR was entirely from 2007 and 2008. Soriano actually had a negative WAR this season (-0.8), whereas Dunn’s was 1.1. For both players, the best season of three was 2007, when Soriano posted a 5.5 WAR and Dunn a 2.9. Would that difference of 2.6 wins have affected the final standings? It’s impossible to say, but the 2007 Cubs finished 2.0 games ahead of the Brewers. While the NLDS sweep that followed might make it a moot point, many fans would regard the 2007 season very differently had the Cubs not won the division.

… and he wasn’t terrible there, though he wasn’t great, either. He played a full year at 3B as late as age 34.

That’s some cynicism.

We will at some point I’m sure discuss the limitations and criticisms of UZR (I hope anyway), so without getting into the weeds, I assume you’d grant that taking a systematic approach to aspects of baseball allows one to at least claim to be “more scientific” than eyeballing something without an overarching strategy? Or do you feel that because there are less scientific aspects to UZR’s analysis (and other saber stats), that it can’t claim any degree of legitimate method/science? (I would remind you that all science, including hard sciences like physics, to varying degrees include elements that rely on “observations of unknown reliability.” Hell, fundamental “laws” related to causality are only true in so far as we’ve never found counter-examples. It seems to me that there are many good criticisms you can make about the usefulness of UZR. But I think that your definition of methodological or “scientific” basis is overly narrow.

• Anyway, as for your points…Given the fact that people still make the claim that defense has little effect on the valuation of a player (see the discussion of Dunn above), it doesn’t seem a large leap to claim that in the past “it was once thought that defense contributes little to the value [plus or minus] of a player,” would it?

• I believe that his usage of “confidence” was in the mathematical sense, whether in the individual case-study or the aggregation of multiple studies. Though Shawn will hopefully correct me if I’m wrong.

• “Treasuring” mistakes is implicit in method based inquiry, which sabermetrics aims to be (which was Shawns whole point, as I read it); without them you can’t better calibrate your study and understanding. Unless you expect every study to actually include the words “I treasure my mistakes!”, then your last paragraph strikes me as pretty uncharitable.