I don't think you can answer whether the 2014 D or 2016 D was better yet. The story isn't over.
What you have so far, is the 2016 D was statistically better in most measures, by a lot in some.
2016 allowed 63 fewer points. That is a lot. Almost every statistic is better in 2016 than 2014.
The counter argument is the 2014 defense played a tougher schedule.
But how do you account for that difference? You are not worse because you play an easier team and do better. The quality of a team or defense isn't defined by its schedule, but by what it does. There is no way to quantify doing better is worse because of who you did it against.
Again its like saying the 2014 defense would be worse than the 2014 defense if it played the 2016 schedule, just because of who it played, even if it played better.
As far as looking at the players, its not crystal clear.
I think the DL is better now. I will take Flowers over Jones today. I will especially take Flowers over the Jones we always saw after Thanksgiving. The depth at DE is also better. At DT 2016 is better. Vince was not prime Vince in 2014, and Branch 2016 is better than Vince 2014 (but not close to prime vince). The #2 DT who played the 2nd most snaps in 2014 was Chris Jones. Enough said.
At LB, its tricky. Hightower and Collins (and lets face it we are a 2 LB team and the 3rd LB plays less than than the 4th DE) are better than Hightower and the combination of guys we are using, but the 2016 LB group has played better than the 2014. Its pretty obvious too, and I don't know why, but somehow the mixing and matching has resulted in a very effective unit.
In the secondary, the safety position is unchanged, and I think the players are at pretty much their same level.
Revis vs Butler favors Revis, but not enormously, but the CB2-4 have been better this year than in 2014. Ryan is playing better, and Rowe has outplayed Browner, especially due to all of the back breaking penalties. People forget Browner was not a full time player.
Ultimately this defense has 2 opportunities to go out on the field and prove how good it is. I think the answer comes after that.