I totally get the argument that one bad game can torpedo a reliever's ERA for while. But Belisle has had multiple bad games, evenly spaced through the season. Capable MLB relievers don't just give up multiple runs in close games every other week. And they certainly aren't capable of "pitching to the score" by only giving up runs when we're already behind.
We should consider ourselves lucky that it hasn't burned us more up to this point and move on. I'm not sure Belisle and Molitor can keep up this jujitsu for much longer.
The question should be what metrics do the front office use. Clutch statistics say he is the best Twins reliever, better than most relievers in the league. Might mean something. Might not . Fangraphs says he has had 10 shutdowns. About a little better than league average . Might mean he is adequate. Fangraphs also says he has 7 meltdowns. Not the highest in the league, but enough to say he might suck. So what statistics does one use to asses how good or bad a reliever is. When Belisle is good, he does the job well and gives the team a chance to win. When he is bad, it is like sending a batting practice pitcher out there. In the bullpen there is Kinzler Rogers and Duffrey. Then there is the rest When they figure out what to do with the spots not occupied by Boshers, Breslow and Belisle (The killed B's) because any of those pitchers really have not fared well. Injuries and regression. Tonkin and Pressly regressed, Chagois, and O'Rourke got injured, as did anyone you thought might have progressed this spring. There are bigger problems than Belisle to be solved.