This idea is basically re-posted from the Ichiro comment thread below, except with sexy visuals and further explanation.
-----
For his career, Ichiro has 315 steals out of 385 attempts, for a success rate of 81.8%. That's an excellent rate that is a good deal better than the league average.
If we keep things simple and call Safeco a 4.5 run environment, then we can value a stolen base at +0.19 runs and a caught stealing at -0.45 runs. Caught stealings have a significantly greater magnitude, of course, because they involve both an out and the erasing of a baserunner. Anyway, using these numbers, then Ichiro's successful steals have contributed 59.85 runs, while his failed steals have contributed -31.5 runs. Add these together and you get +28.35 runs over 385 attempts, or 0.074 runs per attempt.
Since we have eight years of data, let's consider this to be Ichiro's "true talent" value as a basestealer. An interesting question comes out of this: how many additional steals would Ichiro have to attempt per season in order to be worth an additional run? An additional five runs? An additional ten runs?
This is how many:
The dark blue line - the one with the lowest slope - represents how many additional steals Ichiro would have to attempt if he maintained his career success rate. +1 run requires +14 additional attempts, +5 runs requires 68 additional attempts, and +10 runs requires 136 additional attempts.
But is it reasonable to expect him to maintain his career success rate on additional steals? If we assume that Ichiro knows what he's doing, then in theory he's choosing to steal when conditions are best, i.e. when he's most likely to succeed. That means that, if he were to attempt additional steals, he'd be attempting them when the conditions are less favorable, leading to a greater rate of failure. If I put 25 Gobstoppers, 25 orange Tic Tacs, and 25 fun size Almond Joys on a table, and I tell one of my friends to take 20 candies, and then I tell a second friend to take another 20 candies, the second friend will end up with less deliciousness than the first friend, because he had less overall quality to choose from. You could make a pretty convincing argument that the same is true of Ichiro. If he's already taking advantage of his greatest opportunities, then any subsequent opportunities would therefore be worse.
How much worse, I can't be sure, which is why there are five lines on the graph. If Ichiro's success rate on his additional attempts dropped to 80%, then he'd need to attempt 16 more steals to gain one run and 81 more steals to gain five. If it dropped to 78%, he'd need to attempt 20 more steals to gain one run and 102 more steals to gain five. And so on and so forth.
There is the matter of leverage to take into account, which would probably reduce those numbers a little bit. And given that Ichiro's a good basestealer, I think it's safe to assume that if he were to keep attempting additional steals, he'd stay above the breakeven rate, and therefore would contribute additional value to the team. But the total benefit is so very small - 2 runs per another 30-50 attempts? - that it's pretty much insignificant, and not at all a good justification for being critical of our best player. And this doesn't even take into consideration the fact that additional attempts place an additional physical toll on the body, risking injury. We all remember what happened to Ichiro after his hamstring started acting up, and I can't even imagine the tragedy of losing him to a broken finger or something.
There are a lot of people who criticize Ichiro for not stealing enough, or for not diving, or for not being a team leader. These people ought to just come right out and say "I don't like Ichiro," because rather than ascribing disproportionate value to matters of little significance, it's better to simply be honest.