Now, as far as left/right designation and how ratings affect different positions...
Let me first say as much as I wish I could say I can build the new engine to truly simulate everything that goes into a real football play, I can't. There have to be some assumptions made in simulating a play. One of the biggest fuzzy parts of the game is the left/right designation. These are both combined in to "outside", the same as it is in 2.0. To make plays actually specify left or right would take an enormous amount of complexity and require us to open up even more settings. I think it could fairly easily be accomplished in the offensive settings, but when you get to defensive settings, it gets more complicated, and on both offense and defense it would require much more detailed care with the depth charts and formation sets. I don't think we are ready to open up that distinction. What this means in the engine is that it has to account for all "outside" players in an outside rush without assuming it will be left or right. Yes, this means that you don't really have a strong or weak side and you get credit for all of your outside blockers and defenders in each outside play, but the engine knows this and takes that into consideration.
This is the same as it works in 2.0, but the difference in 3.0 is that the match ups are calculated and used a little differently. What I want to see is that faster, more elusive RBs are more successful on the outside and stronger RBs are more successful on the inside. There should also be a different desirable mix of ratings for blockers and defenders inside versus outside. I don't want to get too much into mechanics, but one things that helps this in the 3.0 engine over the 2.0 engine is that the blocking and the tackling is split into different checks. In fact, even tackling is split into two parts - being able to get your hands on the ball carrier and then being able to bring him down. With these checks being split out, we can treat inside and outside runs differently providing a different mix of ratings you want for those players involved. This also holds true for different passing depths for the target and the defenders.
Also we have to keep in mind what it means to be "successful" when it comes to evaluating results. I said I want a faster RB to be more successful on the outside, but players aren't FAST or NOT FAST, they have ratings which are all relative. A stronger RB (higher STR than SPD) might still be successful on the outside depending on the defense he is facing and his blockers. It's tough to say something like "this player would have X number more yards per rush if he had X more rating in SPD". So while it might be tough to see the difference between ratings within a game, and certainly within one play, I hope that we see different players tend towards the stats you would expect within a season. Though even the season are relatively short and could have some variance in results. BUT I'm working really hard to have individual plays fall within certain ranged of expectations based on ratings and match ups and play calls, so that even though one play might hit the unexpected here and there, overall we will tend towards the expected better than what we have with 2.0.
Take for instance the Hands rating in the receiving results. This is a good example of where looking at results over the season is going to tell you more than single plays or games. The chance of dropping the ball should be more consistent based on the Hands rating so we should see more dropped passes with lower rated Hands players (sorry, DIII). However, there are many other things that go into the chance of dropping a pass - how well the pass was thrown, how deep the pass was, and how well the player was covered. There is still a random element to this drop - we can't say "Oh, the player has a 30 Hands so he drops this pass". The goal would be to produce enough results where you want someone to say "hmm, I probably need to get someone with better Hands in that spot" without having a player be totally useless and drop every pass.
So this is probably another post where I'm rambling on and really doesn't add anything to the conversation and will probably be misinterpreted to launch a few rants, but I've been gone for a week, so give me a break. Basic summary of what I'm trying to say is that the 3.0 engine isn't drastically changing how the games are played, but should be changing the results (hopefully within each game and certainly across the season) and control over your settings. The goal being to provide more consistent expectations of how ratings affect the game and more strategy in building your team and your game plans. The downside to the update is that in order to accomplish these seemingly incremental improvements, I need to rebuild the engine because the current structure won't allow for these changes, and in rewriting this code it takes a lot of work, rebalancing and testing to get the new engine to be in the same ball park of results as the old engine.