I just took a look and the weird gap between Xander's UZR and DRS has grown. This has been an annual thing, but in past years it's been a matter of degree; the largest gap between the two metrics was 10.1 in 2017. UZR has consistently seen Xander as an average-ish shortstop, maybe just a hair on the fringey side of the line, while DRS until this year saw Xander as a poor but not downright awful defender.
This year, DRS says Xander is the second-worst defender in the American League, topped in awfulness only by Miguel Andujar. According to their calculations, Xander has lost the Sox almost two net games with his glove. In the meantime, UZR continues to say that Xander is essentially average. The gap between their two estimations is a thoroughly ridiculous 20.8 runs.
I'd be really curious to see an explanation of how two systems can look at the same set of events, using superficially pretty similar processes, and come up with that pair of answers. I know some of the differences between the two methods -- among other things, DRS uses smaller buckets and compares to a one-year average, UZR large buckets and a three-year average. But it's a bit mind-boggling that what seem like relatively minor methodological differences can produce such enormously different results. The gap between the worst DRS and the best UZR in baseball this year is 45 runs -- and the range of disagreement about Xander's defensive performance occupies nearly half of that gap.