I think people that don't understand this really just don't understand how a basketball game is won (# of possessions * efficiency of each possession).
...
The whole point is, there's a colorable argument here that even if you factor in defense, JG is merely an equivalent player to Bonner. And Bonner is essentially a role player. Hence how one might come to the conclusion that JG could not, under any circumstances, be even the fourth best player on a championship team.
Of course the people who don't understand stats or what I'm saying will just repeat themselves - "Stats can show anything!" (actually no, they can't). Or they'll just claim that what I'm saying is preposterous and can't simply be true "Come on, you don't REALLY think that Bonner is better, do you? You're obviously insane." (despite the fact that Bonner has a higher TS%).
I'm pretty sure it's you who doesn't understand how to interpret the stats. This is a bleeding example of confirmation bias - you took a hypothesis (Jeff Green is a low efficiency player who would be 5th best or worse player on a championship team), and went out to find numbers that supported your opinion.
Even if you just focus on win shares, which is a nice metric but not the only one out there, you have to at least consider total win shares and the team they were on.
When SA won the title in 2006-2007 they won 58 regular season games over which Bonner was 10th on the team in total win shares (2.2).
By contrast, in Jeff Green's first 2 years in OKC he contributed 10.8 win shares to teams that won a total of 73 games.
By that measure, Bonner contributed 3.8% of his team's wins, and Green contributed 14.8%. If you gave Bonner 3x the minutes AND his efficiency stayed exactly the same playing 30+ minutes against starters and bad matchups for him, he still wouldn't carry as big of a share.
So yeah, I didn't even need to change metrics to go find a straightforward argument that Green is better. See how numbers can be used to support an already formed opinion?