Okay, folks, let's clear the boards here, take a deep breath, relax, and look at this from a slightly different perspective.
Let's start this off with linear weights. I like linear weights. I like them a lot. So when I refer to runs throughout this piece, know that I figured out these runs using linear weights -- specifically Batting Runs. A full examination of the process involved can be found here. To convert runs to wins, divide by 11 (it's a rough estimate, but it works for this estimate so long as I stay consistant.)
I need to clarify here - Batting Runs are baselined, so that negative Batting Runs are possible. Taylor Green of the Milwaukee Brewers, hitting .249/.311/.376 according to his 2008 CHONE projections, is our 0 offense player. Looking at the lineup simulator, a hypothetical 0 offense team - 9 Taylor Greens - scores 3.9 runs a game.
I cannot find the results of the last major discussion we had on this topic where we ran all kinds of lineup simulatons; if anyone can find it I would be grateful. But sumarizing briefly, using the assumptions of the lineup simulator, we got something between 1-4 marginal wins by shifting Soriano around in the lineup. Of course, we ended up batting the pitcher eighth LaRussa style, using Fukudome or Lee as our leadoff hitter, and batting either DeRosa or Soto in the three hole, so it's not like these were likely lineups.
Okay, but here's the dirty little secret that the lineup estimator is keeping from you: the number of plate appearances by lineup position.
Those figures are for the 2007 Cubs; your milage may vary.
Using Batting Runs (and the CHONE projections), it looks like Soriano produces 0.061 runs per plate appearance. Batting leadoff, that's roughly 47 batting runs, or 4.27 wins. Drop him down to sixth; 42 runs, or down to 3.8 wins. Then let's sub in Brian Roberts; he produces 0.044 runs per at-bat, or 34 runs as a leadoff hitter. That's 3 wins. Drop him to sixth and he produces 30 runs or 2.75 wins.
So: Soriano 1st + Roberts 6th: 80.84 runs, 7.34 wins.
Soriano 6th + Roberts 1st: 72.07 runs, 6.55 wins.
I didn't bother calculating out Soriano's lower Batting Runs for his lower career splits batting out of the leadoff spot. So reevaluate accordingly, if that matters to you.
There are hitters better than Soriano; we even have a few of them on the team. But they won't be batting leadoff, sadly. And Soriano is a better hitter than Brian Roberts. A much better hitter. And if you want to win ballgames, you need to give more at-bats to your best hitters. And that's what you miss out on if you simply look at the lineup as a per-game construct, rather than how it plays out on a full season.
The problem is that there's way too much of an emotional attatchment to the "Traditional Leadoff Hitter." Of course, the Traditional Leadoff Hitter is an artifact of a time period when run socring was a lot lower, and so a stolen base was a much more valuable play than it is today. And then we have the people worrying about RBIs, as though the "right" people getting credit for driving runs in was more important than putting actual runs on the board.
Everybody, learn to quit worrying and love the swinging strike. It's for your own good.