It's more intimidating in motion, probably.
For the last couple of months the St. Louis Cardinals have had two former relievers in their mid-twenties in the rotation, and I'm going to be honest: I didn't expect there to be much in the way of controversy when the second one, Joe Kelly, was moved to the bullpen to make room for Jaime Garcia. This is just another example of how out of touch I am with the common man, which I encourage you to use against me if we're ever running for the same political office: I failed to realize how much people loved Joe Kelly, and how little they weighed Lance Lynn's first two months (and July.)
So: Quickly, here are some reasons why other people believe the Cardinals should have stuck with Kelly that explain why I believe the Cardinals were right to replace him.
Look what they've done over their last 12 starts
I think it's safe to say that Joe Kelly's last 12 starts—the entirety of his MLB career—have been his best ever. His FIP is 4.03, give or take. (I'm calculating both of these manually, since I can't get this next number on FanGraphs.)
Lance Lynn's last 12 starts begin with his 11 and 12-strikeout performances against the Astros and White Sox but move on to include all but one of the starts in which he's allowed more than three earned runs. His ERA over that span is 4.65; his FIP is 3.93.
To get Lance Lynn's FIP over Joe Kelly's you have to line up your endpoints around his last four starts (5.11), ignoring the 19 innings in which he'd allowed a total of one run directly before that (3.83), or push past that to include the game in which he allowed three home runs but not the two 10-strikeout games immediately preceding it.
Surely Lance Lynn must be tired
Lance Lynn has never thrown more than 138 innings in a season, aside from 2010, in which he threw 164, or 2009, in which he threw 148. Joe Kelly has never thrown more than 141 innings in a season.
Lance Lynn has allowed a ton of hits
Which is to say that his BAbip over his last 12 starts is .349, over his last 10 (the first slump-only arbitrary endpoint) .354, over his last four .369. He has allowed a ton of hits.
Joe Kelly, to his credit, has himself allowed a fair number of hits (.318.) Not quite a ton, though.
I got a tweet a few days ago suggesting Lynn was putting this team in a hole every game, and that that was a particular problem worth addressing (by leaving Joe Kelly in the rotation.) Joe Strauss helpfully names the first innings that have been a problem—he's allowed three earned runs in three of his last four starts. That's nine earned runs in four innings. Unfortunately, he also uses the full-season stat (an 8.22 first-inning ERA) right after, which seems a bit misleading—outside those four starts it's 5.68, which is bad but also says a lot about how small the sample we're discussing is.
Here's what's happened in those three particularly bad first innings:
August 12: He hit Juan Pierre; he allowed a Kevin Frandsen bunt single; he allowed a triple to Chase Utley and a grounded single to Ryan Howard. The real problem here is the line-drive triple to Chase Utley, on a 2-0 changeup. Before that he had Pierre 0-2 and Frandsen 0-1.
August 7: Angel Pagan and Marco Scutaro hit ground ball singles past second base (Skip Schumaker.) Melky Cabrera (RIP) grounds out to second next, before Buster Posey hits a long home run to right-center. Hunter Pence and Brandon Belt pop out.
I watched this just now and Lynn just misses on a 2-2 fastball to Pagan before allowing a ground ball that is a reasonable argument for playing a second baseman at the second base position more often; Schumaker has to dive for it, and gets a glove on it, but I think somebody else might have played it standing. Scutaro, on a 1-2 fastball up, gets a genuine single on a grounder through the hole to right.
July 27: This is the legitimately terrible inning. David DeJesus triples, Starlin Castro hits a hard single into center field, and Anthony Rizzo homers to drive in Castro. Lynn allows another walk and a double before getting out of the inning; he'd have allowed four runs if Yadier Molina hadn't caught Bryan LaHair stealing(!)
These are bad, but only the July performance seems easily transmuted into a genuine issue; the other two are big hits mixed with ugly sequencing of outs and near-outs. In fact, it's instructive to look at the first inning in which he didn't allow any runs, on August 2—he allowed a double then, too, it was just to the leadoff man, and followed immediately by two groundouts and a flyout.
It could be that Lynn has a problem with first innings, but the first innings we've seen, all 23 of them, aren't nearly enough to make that diagnosis. Drilled down, a lot of them show the usual vagaries of baseball rallies, and meanwhile it isn't especially hard to find somebody else with terrible first inning splits one year and nothing the next. (Adam Wainwright, for what it's worth, has a 4.88 ERA in first innings this year.)
If there's a problem with Lance Lynn—or a thing worth worrying about that could become a problem—it's that nothing in his minor league performance to date suggested he'd be capable of a season like this. That's even more true of Joe Kelly. And that's why Joe Kelly should be in the bullpen, and Lance Lynn the rotation.