Oh and by the way Millage has got to be nearly as bad as Dunn. All 7 of our first games this season Millage lost the game with a blown play in the outfield at least once. My memories of him are Millage running in semi-circles while the ball flew over his head repeatedly. And he always grounded out on the first pitch if someone was on base. Pitsburg could not have been dumber with this trade!!!!!

It can make a significant difference in a player’s value. Michael Cuddyer has a great arm and awful range, so just using range metrics makes him look like one of the worst OF in the game (instead of “slightly below average”).

While the metrics used here suggest that they have equal value within the Nationals’ system, could there also be a positive effect on Dunn’s defensive value if he has an elite fielder in center field to cover some of his poor defensive ability?

Even if it is a small difference, Dunn may be relied upon less to make plays in the field. This, in turn, could make up for at least some of his negative impact on defense. While in turn, it’s possible that Morgan’s defensive measures will decrease with his added responsibility on the field. Morgan had the benefit of playing alongside an amazing defensive CF for a few weeks in Andrew McCutchen in Pittsburgh, and while Nate McClouth was measuring pretty poor defensively himself, he at least had the speed to cover ground in CF as well. The downgrade in the defensive ability from McCutchen/McLouth to Dunn is significant.

I’m not an active follower of sabermetrics, and I suppose this is more of an intangible argument than something that would be covered by sabermetrics, but I believe it is something to consider.

Dunn’s consistent lack of contact combined with patience and power make him an above-average offensive player. His abject inability to play passable defense at any position undercuts his overall value severely. Frank Thomas during the first 10.5 seasons of his career, a point which Dunn is fast approaching, was as good as Albert Pujols. Think about that before you make such a ridiculous comparison again.

I tend to agree about the undervalued-ness of defensive metrics, but I think you may lean too far the other way here. I tend to agree with the commenter who said that the confidence in and proven utility of the metrics should factor in to how you weight them. The offensive metrics have many more years of proven utility and proven predictability. Defensive metrics still suffer from lots of noise. Regressing Morgan’s defensive metrics for sample size is necessary, but I think the noise in the metrics may also argue against using a direct offense v. defense comparison in this way.

Moreover, this also brings up one of my main methodological misgivings about both the offensive and defensive metrics. I know both are denominated in “runs” but they sort of measure different things – the offensive metrics measure runs created, and so can be “normed” to the overall level of runs scored in a given year – the size and scope of the contributions can be kept within the overall context. But there’s really no way to measure, in the aggregate, how many runs are prevented each year – so there’s no context for whether a run prevented is really the equivalent of a run created offensively.

So, that calls into question whether Morgan’s +35 defense is really only ten runs worse than Dunn’s +45 offense. 22,585 runs scored last year. How many were prevented? Without that comparison, we can’t really compare offense to defense this way – the underlying assumption that a scored run = a prevented run may or may not hold. It may actually be that Morgan’s +35 > Dunn’s +45, or that Dunn’s +45 is worth +90 runs defensively. The comparisons are always runs above average, but that average tends towards the mythical, since it’s not tied down in the reality of actual production, leaving open to question the comprability of the numbers across players and offense to defense.

The difficulty, as Dave says, is in comparing players at the extremes. A team of all Dunns would undoubtedly put up great offensive numbers, but be awful in the field. A team of all Morgans would be equally ineffective the other way. I think the point that sabermetrics tends to value the more easily measurable (offense) and discount the more difficult, less confident more controversial (defense) is a good one, but I think you may have leaned too far to try to prove a unlikely supposition.