The computerees at Baseball Prospectus are revamping WARP, or Wins Above Replacement Player. WARP is like Wins Above Replacement, in that it tries to answer the question, "How many games did Miguel Cabrera (or whoever) win for his team?" But it differs from the Baseball-Reference and FanGraphs variants in how it's calculated. The biggest difference, I think, is that the Prospectus guys are very skeptical of defensive statistics based on "video scouting," where a bunch of guys watch every ballgame on TV and plot (or estimate, depending on your point of view) the location of every ball handled -- or mishandled -- by the fielders. BP have their own fielding stats, which are based on plays made ... "Okay, this shortstop made X number of assists in Y innings in the field, and lefty pitchers accounted for Z innings, and the staff's groundball-to-flyball ratio is N, so we would expect 250 balls hit to him ..." I'm oversimplifying, but I think that's the gist of it.
Anyway, late in the article, Colin Wyers says this:
One commonly proposed test of how well a total value system does its work is looking at how well it predicts team results. This is dangerous ground for sabermetricians to stand on; it turns out that old-school stats like RBIs and pitcher wins do a much better job of that than any total-value stat proposed. The entire point of this exercise is that we are willing to sacrifice the best possible accounting of team wins in order to do better at expressing an individual player’s contributions, isolated as best we can from those of his teammate. Using reconciliation to team results causes us to lose sight of that goal and provides very little insight into the quality of our work.
I'm not sure what Colin means by "old-school stats like RBIs and pitcher wins do a much better job ... than any total-value stat" at predicting team wins. Sure, pitcher wins predict team wins perfectly, but that's a tautology. RBIs are a good, though not perfect, predictor of runs scored, but again, that's really just accounting.
What's more, if one of the goals of the creators of total-value stats is to encourage their adoption among baseball fans, why not reconcile them to team wins? Bill James' Runs Created does this very thing. If, after calculating the Runs Created for every player on the team, there's a discrepancy with the actual number of runs the team scored, the Runs Created totals for the individual players are adjusted up or down in accordance with their playing time. Win Shares, of course, starts with team wins, and apportions credit first to the offense and defense, then to individual hitters and fielders.
I suppose that's a philosophical difference, and I'm honestly not sure who has the better argument. It's not as if Win Shares caught fire with the public, after all, though it was synchronized to team wins.
I'm glad Baseball Prospectus is working to improve WARP, and I hope someday we'll get a thorough reworking of Win Shares, too. And maybe we shouldn't worry too much about their universal adoption. At least not just yet.