We could spend hours trying to figure out what Zach Duke did in 2005 that he hasn't done since then and what that might have to do with his success that year and his lack of success since then, but Duke's story is an easy one to figure out: he doesn't just doesn't miss enough bats. In his rookie year, Duke struck out more than 6 batters per 9 innings, since then his K/9 rates have been 4.89, 3.44, and 4.23. Check out the K/9 leaderboard for qualified starters in 2006, 2007 (Duke didn't qualify due to his injury, but you can see where Duke's 3.44 would fit), and 2008.
That said, Duke was really a lot better last year than his 5-14 record suggests. Many of his peripherals match up pretty closely with his 2006 season, which everyone mostly agreed wasn't great, but wasn't exactly unexpected. The reason I'm pointing this out is because I think a lot of people expect the Yankee trio to boot Duke out of the rotation this year and I don't think that that's a given by any means. In both 2006 and 2008, Duke's batting average on balls in play against was .327. That's a good 30 points higher than the league average and a lot of that can be contributed to the Pirates' defensive failings. Using the same PMR conversion to runs that I've been using, Duke lost more than a tenth of a run of ERA to the Pirates' defense this year.
What always gets me about Duke is that he's supposed to be a finesse pitcher, but his control isn't all that great. His strikeout rates are always low, but his walk rates aren't particularly low to match. The only season when he's had a K/BB ratio better than 2:1 was that rookie year. I think that it's certainly possible for a pitcher to be an adequate starter while striking out less than 5 hitters/9 innings, but if you're allowing hitters to put that many balls in play, you've got to minimize the number of free passes you're giving out.
It's hard to write a lot more about Duke, because he's really one of the more scrutinized pitchers in the rotation. I will say that people (myself included) do tend to way over or underestimate Duke's actual value. There's one crowd that finds him completely worthless because of his terrible record last year and his awful 2007 season. There's another crowd that still thinks he can find the 2005 magic and turn into an ace again. I'm quite guilty of this as well. I mean, I just wrote a bunch trying to sell you on Duke not being as bad as he pitched last year, but where did I mention the 646 hits he's given up in his last 507 2/3 innings? Duke can't be more than a fourth or fifth starter until he stops doing that.
If we give rotation spots to Maholm, Snell, Gorzelanny, and Karstens (which Neal Huntington won't do right now, but I think it's a safe bet), then Duke's probably in a battle for the fifth spot with Ross Ohlendorf right now (unless Braden Looper walks through that door). Would the Pirates be better suited with Ohlendorf in a set-up role and Duke in the five slot, assume Duke can bring the strikeout rate up a little more and the defense can help him out some? I think they probably would, but then I'm not sure I'd gamble on Duke being a lot better than last year, either.