In case you've had your head under a rock lately, there's a raging debate going on right now at this site about pitcher performance and how to measure it.
Let's start with an assumption. At the end of the day, "performance" for a pitcher is measured in runs allowed. That's the only stat that actually affects the team. Nobody would give a crap about Yusmeiro Petit's outrageous HR rate if it didn't affect runs.
ERA is better at measuring the runs a pitcher has allowed in the past than FIP. Is there a correlation between runs and FIP? Of course. FIP is easily calculated, better at isolating a pitcher's performance in the things he can directly control, and has decent run prevention correlation, which is why it's used. But ERA is better at conveying his "performance". FanGraphs has taken the opinion that defense behind a pitcher is irrelevant, as are balls in play, since there are so many other factors that affect them, most of which are well outside the pitcher's direct control.
You can't quantify "feel", but FIP just doesn't always feel right. Zach Duke's 3.65 FIP doesn't "feel" as accurate a measurement of his performance as his 5.28 ERA. And, let's be honest -- Joe Saunders' 4.80 FIP doesn't "feel" as accurate as his 3.56 ERA.
But why is it that we (meaning, the more Sabr-oriented folks) are more quick to dismiss Duke's outlier performance than Saunders'?
On the other hand, just because some guys, like Joe Saunders and Matt Cain, always seem to outperform their peripherals in the "runs prevented" department, isn't reason to reject the entire study of underlying peripherals. Just because Duke always underperforms his FIP doesn't make FIP, or any other complex sabermetric stat, irrelevant. However, a trend like that does seem to suggest that there are problems with looking entirely at the parts of a pitcher's performance that FIP examines.
I think everyone here can admit that FIP has plenty of flaws, as does ERA, especially in the uneven rules regarding what constitutes an "earned run", and especially as it regards different parks and leagues. ERA+ attempts to fix the latter, while FIP attempts to fix both.
All this means is that we need to find a BETTER way to quantify performance. It doesn't rule out the value of so-called statistical analysis. NOBODY rates guys by their gut anymore without some kind of number to back it up, and I don't think anybody here has tried to defend Joe Saunders' performance to date without conjuring his impressive ERA. The thing we're debating is, what's the best way to quantify a pitcher's ACTUAL performance?
Let's pretend that Joe Saunders works as a waiter and makes $5 an hour. If he's checking his checking account balance, he doesn't completely ignore tips, even though tips are largely luck-dependent. Getting a $200 tip from one wealthy customer who's in town for a convention one night shouldn't be predictive that he'll get an extra $200 tip on this day next week. He should budget for next week based on what he typically makes between tips and salary. At the same time, that $200 is money in the bank.
FIP, on the other hand, is like taking ONLY his salary next week into account, and assuming he'll make whatever an average waiter, across the entire Major League of waiters, makes in tips. Is tip money entirely within his control? Of course not. But let's say that Joe Saunders is a young, attractive female, who's able to flirt with the hitter and convince him to offer at a pitch out of the zo- errr, able to charm a customer into leaving a slightly better tip. Zach Duke, on the other hand, goes about his job fairly well. He gets food out to his customers immediately, he's always on time, and he never gets an order wrong. He's also surly, lacks a personality, and has trouble connecting with his customers. Duke typically earns less than the average waiter in tips, despite his excellent underlying waiter peripherals.
That difference between "tips" isn't taken into account by FIP, which assumes that all customers are the same, and ignores other important waiter attributes like charisma. Joe Saunders may have heavily outperformed his FIP so far this year, and it's not unreasonable to expect his ERA to go up over the rest of the season.
But the run prevention he's performed so far is money in the bank. He's not giving that $200 back.