I inadvertently omitted Olynyk from that list, but the case is simply that based on whatever plus/minus metric you want to look at (e.g., RAPM, PIPM), Sullinger, Olynyk, and Johnson in particular graded as strong contributors (generally between +0.5 and +2 point per 100 possessions better than average). For Olynyk and Amir, that's been true on other teams as well, so it's not just a question of Stevens' usage.1) Why? And I’m sincerely curious. I know that you play with advanced metrics more than I do. What is the case that the 2015-2016 versions of Sullinger/Johnson/Jerebko/Zeller are “markedly” better?
Meanwhile, Kanter is somewhat reliably among the worst graded players in the league by the same metrics. While Theis has actually graded well in limited PT, it's not many minutes, and in a limited context. Meanwhile TL and Poirier essentially have no data. We'll see how Stevens uses them - who knows, maybe he gets water from a stone with Kanter - but at present, I wouldn't bet on it. The net result is that the 2015-2016 team had three strong bigs in Johnson, Olynyk, and Sullinger, while the current team has maybe half of one with Theis (and a couple TBDs with TL and Poirier).
I know this isn't a totally satisfying answer ("they're good because the mostly black box data says so"), but I mostly believe it in this case.
That would change the assessment of the bigs, yes (although obviously weaken the assessment of the wings).2) Does including 32 minutes of Tatum as a de facto member of the group in a smallball / singlebig deployment change the calculus?