It appears as though calling the proper defensive formation is starting to have the expected impact. I've been testing this through sample games, but the first two games of my Craven season are indicative of what my testing has shown - guessing wrong against the offenses is much more impactful than in 2.0. 

Game 1 - ran a standard defensive gameplan in the first half, 5-2 balanced on 1st down and on 2nd/3rd on medium/short, nickel on 2nd / 3rd and long. My opponent was set to basically all run, including trips. Here are the RB stats:

Player	Pos	Rush	Yds	Avg	TD	20+	St
Alan Freeman	RB	33	152	4.6	1	0	6
Robert Collins	RB	3	1	0.3	0	0	1

2nd half, I shifted the game plan to all run and 5-2 in every formation / down / distance. THe RB performance dropped off significantly:

Player	Pos	Rush	Yds	Avg	TD	20+	St
Alan Freeman	RB	28	82	2.9	0	0	9
Robert Collins	RB	7	18	2.6	0	0	2

Game 2 - kept the all run defense from the 2nd half of game 1, 5-2 vs. most formations / down distance, and the opponent RBs were shut down:

Player	Pos	Rush	Yds	Avg	TD	20+	St
Christopher Hargrove	RB	10	25	2.5	0	0	3
Gilbert Johnson	RB	6	11	1.8	0	0	3
Paul Parrish	RB	1	16	16	0	0	0

2nd half, I shifted to dime / all pass against ND Box, just to see if, as we see in 2.0, a coach who forgets and leaves a passing defense against an all run team can see success. The answer is a resounding no, not just in this game but in overall tests.

Player	Pos	Rush	Yds	Avg	TD	20+	St
Christopher Hargrove	RB	12	144	12.0	1	2	2
Gilbert Johnson	RB	9	85	9.4	0	1	1

Small samples, but it has shown up for me in testing. It's no longer a set and forget game, and the importance of gameplanning may have been resurrected.

I agree.  I can see much more cause-and-effect in gameplanning.  

The only thing I don't like about that is that, as someone else posted earlier, in real life if a team is rushing the QB out of shotgun every play the defense will adjust in real time.  Here we only have one opportunity to adjust settings and even then the other coach can do the old switcher-oo.

My suggestion would be to limit the effectiveness of the non-intuitive gimmic offenses if possible.  That is limit the effectiveness of a wishbone passing scheme, even if the defense is all-rush, or limit the effectiveness of a shotgun rushing plan even if the defense is all pass.

Nobody wants this to turn into a heads or tails game based on guessing the right defense for a gimmic offense.

Maybe make the defense vs offense call only for the set-up of each play. In real life, before the snap the defense is just guessing what is going to happen, but then they adapt when they see if it is a pass or run. In the GD world, make that the same - for the first two levels of game decisions (at the line) the O pass vs D rush (or O rush vs D pass) has an advantage, but after that not so much.