And that seems to be the core of the argument against WPA. "What if something else had happened?"
As the game actually played out, the Twins scored a run in the top of the first, for which the batters got suitable credit (surely you agree that scoring the first run in a game brings a significant advantage), and then Ober gave away that advantage with a home run in the bottom of that inning. The hitters put the game away, for all intents and purposes, in the top of the second. Ober still gave up a home run in the bottom of that inning too, which was much less damaging, and from there on he was lights out - but by that point they would have likely won (96 times out 100, with that lead, according to b-r.com's tally) even if he or some relievers who followed had been less effective.
Bailey Ober had a fine game. But the hitters won it, and during the brief period the game was still in doubt, Ober was not effective at the job of holding the lead. "What if those hitters hadn't scored in the top of the second? What if the big inning had been in the top of the ninth?" Different, non-existent ballgames, those. It's not Ober's "fault" that he had little opportunity to affect the winning outcome, but that first inning wasn't a plus.
I'm not one to dwell on "clutch" performances too much, but to whatever degree major leaguers are clutch, WPA may reflect it. Joey Gallo last year had a negative WPA despite positive WAR, because the stats that feed WAR seemed to come only when the team was already comfortably ahead - and that seemed to agree with the eye test of all of us who disliked having him in the lineup. I'm not going to cast aspersions at Ober now, but merely point out that in this one game, his performance got better when the game was not tight anymore; other games, Ober has done well when it counted.
Pablo pitched his masterpiece on Sunday under much tighter circumstances, clinging to a 1-run lead for six of his eight innings, and did better by surrendering zero runs, not two. WPA reflects all of that too. The batters other than Buxton (and Castro and Correa to a degree) didn't win that game, Lopez (to whatever degree we assign anything to one person) did - three runs by the offense loses more times than not.
One final observation, WPA might reflect the fan's mood as the game progresses. I enjoyed my afternoon in the sun in Oakland in a much different, more relaxed, way with that early lead than if it had been a tight ballgame. The much-maligned Manuel Margot gave me that luxury as a fan, and he deserves the WPA he earned.
WPA reflects the game, as it actually went, better than the box score and rhetorical arguments do. Does the stat have flaws? Yes! Is it the one ultimate stat that tells us how good a player is? Not at all! But criticizing it for not doing things it's not designed to, isn't any more fair than me criticizing RBIs because they don't include defense.
Someone else called it a garbage stat, not you, I believe. That's no way to start a reasoned discussion.