Some relevant statistics re: Koji's (non) usage:
- the Leverage Index in that situation (bases loaded, two out, tie game) was 6.41. For some context, an average situation has a leverage index of 1.0 and the Leverage index in the top of the ninth was 2.14. That would have been the second highest leverage situation that Uehara faced
all season.
- The Red Sox had a win probability of 35.1% at that point. This is probably high, as an out gives them a 50% chance (new inning, tie game), implying that they had a 15.1% chance of losing the game there in the 9th. Marco Scutaro, even against a top closer (and certainly against Villareal), would have an OBP over 0.151.
- If Scutaro gets on, the Red Sox lose (win probability of 0%). If he is retired, they are back on even footing, 50-50, as stated above. Thus, the swing value across both relevant results from that plate appearance was 50% of a win. That qualifies as fucking gigantic by any reasonable definition.
- If Uehara were to retire Scutaro and be lifted for a pinch hitter the following inning (not required, as the Sox had two double-switch options available in Ortiz and Salty), his WPA for that outing would have been +.149 and would have been his 9th most impactful appearance of the season. Of the eight more impactful outings this season, he pitched two innings three times and recorded only two one inning saves.
- Uehara had not pitched since the 17th and was sufficiently in need of work to close out a 12-1 blowout the next day, literally a zero leverage situation. It seems extremely likely that he could have pitched more than a single inning (certainly a single inning plus one batter) if necessary.
Additionally, the (sizeable) gulf between Uehara and Villarreal is significantly larger in a situation where a baserunner instantly wins the game than it would be starting out an inning (should it come to that).
I don't think Farrell is an idiot or tried to throw the game, but he absolutely made the wrong call here.