Boston needs Tatum. Boston doesn't need Brown.
I’m pretty sure we need both.
Except Boston is actually better with Brown on the bench and has been for the last 3 years. That was the context of that quote you pulled out, without actually commenting about it. Brown's value to winning is basically non-existent. He has great stats, but he doesn't impact the scoreboard and never has, which is what matters at the end of the day.
Sure wish you could've bought into this type of argument in our Moses/Kareem debates where +/- metrics far better than raw on/off were used .
As for Brown, adjusted +/- metrics (eg. RPM, RAPTOR, multiyear RAPM) that paint a far better picture of a player's situational value than raw on/off suggest that he's a small positive on both ends of the court - in line with a fringe All-Star when you consider that these stats generally underrate his ability to guard a variety of wings or even guards in the playoffs (for example, RAPTOR thought that he was a neutral defender in the RS last season, but they rated him as a top wing defender in the playoffs). The big thing about Jaylen this season is whether his qualitative (eg. shooting, passing, dribbling, etc) and quantitative (in terms of the box score) improvements can actually drive team success like a borderline All-NBA guy - we still don't have adjusted +/- data yet, those take 20+ games before they are released, and raw on/off in a 7 game sample size might not even pin the directionality of a player's situational value correctly (you need a 20ish game sample size to even start getting a reliable-ish signal).
They don't have on/off differential available back when Moses and Kareem played, but I suspect that Moses had a larger one because he was much more important to his team's offense than Kareem was in the the late 70's and early 80's and that affects the scoreboard a great deal.
Raptor often has some pretty ridiculous results, but that said their on/off stats show exactly what I said, that Boston is better when Brown is on the bench then when he is in the game, while Tatum is by far the most important player on the team and that Boston falls apart when Tatum isn't in the game. Last year, Tatum's on/off Raptor was +7.1. Hayward was next on the team at 3.9. Brown was -0.8, only Semi was worse of the 10 Celtics that logged enough minutes to be tracked.
And to be clear, I'm not suggesting that if Boston just removed Brown they would be better, but I don't think the loss of Brown would be felt nearly as much as this board. And even last year, Brown missed 15 games, the Celtics had the exact same win percentage in those 15 games as they did over the season (66.6%). Tatum missed 6 games, Boston was 2-4 (33.3%) without him and 46-20 (69.6%) with him. And this wasn't just last year either. In 2019, Brown missed 8 games, Boston went 8-0. They went 8-4 without Brown in 18 (less than a percent worse). The simple reality is, for as good as Brown is, the team just doesn't miss him when he isn't there.
There are so many issues and problems with APM analysis that it isn't even worth getting into it here. Suffice to say that the error bars on these sorts of analytical methods tend to make them worthless even if they are sometimes wrapped broadly, like giant cushions around an actual valid conclusion.
IF there is any validity to the contention that the presence/absence of Brown has been of lesser impact then you can probably find the actual reason for that in utilization and touches.
A player not being utilized on offense is going to have a lesser impact than a player getting lots of usage. That's just fundamental.
Jaylen Brown has historically been somewhat scarcely utilized. This year so far represents a notable jump from prior seasons:
Touches/gm
Player 2016-17 2017-18 2018-19 2019-20 2020-21
Jaylen 21.9 38.9 31.8 49.0 59.1
Jayson 42.7 45.3 68.1 74.4
Now, starting in the Spring of the 2017-18 season, Jaylen's scoring efficiency started to increase and it really kicked up starting in late November of 2018. But his touches went down because of his ice-cold start to the 2018-19 season. The overall impact of his offense was still muted. Even during some of his good playoff runs (2018, this last year), where Jaylen got more shots (higher USG), his total touches were still relatively modest.
One good sign though, is that during stretches where his USG was higher, Jaylen's scoring efficiency (TS) did not drop. He was always efficient regardless of higher load.
This season (so far), Jaylen's touches and USG are both up significantly over prior seasons. And his efficiency is (so far) definitely holding up just fine. And you can already see the advanced stat results of the higher utilization in things like his Win Share rate, which is over .200/48.
If these trends continue, I suspect that Jaylen's value over replacement will be a quite a bit larger this year than in prior years, which would render your analysis moot.