Adaptive AI - Play Based vs Route Based

[Valdarez]

First, I have to say that Adapative AI is something that competitive football games have needed for a long, long time. There are 11 guys on the field, and they should all be playing based on instinct and experience in combination with the play calling from the player.

User Feedback (Route Usage)
The Adaptive AI that Ian described in his blog appears to only be routed based. I'm curious how the development team will help the user know what routes he has thrown to so he doesn't accidentally fall into the trap of using the same route over and over again?

Play Based vs. Route Based Adaptive AI
The current implementation appears to be based on the routes, instead of the play. I wonder why they don't do it based on both? Give the defense a huge bonus against the 4 to 5 play wonders out there.

User Feedback (Play Usage)
It would be nice if they somehow conveyed the number of times you have called a certain play. I don't know about you guys, but I tend to forget what I have called by the end of the game, and while I could write the plays down, it would be really nice if the game kept track of them for me.

Keep track of every play a team plays on offense. Every time a play is called, it goes from green to yellow shading. Then the next time it's called, it goes from yellow to red shading. The color indicates how well the defense will recognize the play. The defense will get a speed / recognition boost so they can better lock down the play, even though a player might not have called the right defensive play. This would help avoid scenarios where players are able to be effective using a limited set of plays all game long.

They could add a 'cooling' effect based on every quarter or half. I would suggest half, so a team can effectively have no penalty what so ever for calling the same play twice a game.

This is a really bad mockup, and uses APF2K8's graphics, but at least you can get an idea of the defensive awareness bar at the bottom of the play, or you could just shade the entire area a given color.



If nothing else, you could just have a little number in the corner that shows how many times the play has been called. It would just be nice to have something that showed how many times a certain play has been called, and how likely the defense is to pick up on it, with the latter being more important if route based adaptive AI is in play.

[Bmlives]

Adaptive AI is adjusts to Plays and Routes. The Blog, however, just sounded like it was Route-Based because Ian spent most of his time explaining it in a broad manner.

"For example, if an offense is running the exact same play over and over again for big gains, we would boost the defense’s play recognition ratings to emulate them learning for their mistakes to have a better chance to react when they see the same look again. Same goes for something like continuous out routes…"

- Quote from the Blog

[Valdarez]

I think it's going to be very important, from a competitive standpoint, that if you are going to be penalized for a certain behavior (running the same route multiple times) that the game give you some type of warning / notification so you can avoid the penalty. I'm old, and my memory isn't what it used to be. I'm lucky if I can remember the play I'm running, much less the 20 plays that came before hand.

[Bmlives]

From what I've come to understand, it only picks up things you do "Successfully".

So, I'd expect people to remember if they can constantly throw the deep corner routes with success, or get back to back to back first downs off of comeback routes.

Either way, I can understand that people might be unaware of things they run over and over, but you shouldn't have much to worry about if you just mix-it up without knowing it. I just call random plays. But eventually, even if you switch it up within a gameplay, if you have success with different things, they can never really pick up on ONE thing... So actually, in some cases, Adaptive AI might never even show up in a game.

[Sven Draconian]

Firstly, this better not be a factor if you run a play 3-4 or even 5 times. There really aren't that many "different" plays and having any sort of offense means you have a base package that you run.

If the adaptive AI starts to kick in the 3rd time I run a play then it's going to be an awful feature (IMO). Assuming a person can run 30-40 offensive plays a game it should be no fewer then 5 plays before the adaptive AI starts to turn on.

I would also support a cooling element. Just because I hammered you with a certain running play in the first quarter doesn't mean you'll remember in the 4th if I stopped using it.

With that factor in mind, I don't think there should be any warning. That would seem to defeat the purpose of it. You became predictable, so the defense is predicting your plays. As long as you don't constantly run the same play, you should be fine, assuming 35 snaps are run, calling a play 6 times puts it at 18& of your offense. I

***
I really hope this isn't overpowered. A boost in AWR and play recognition would be great, but execution should still matter. Denver ran the zone play 20+ times a game and it still worked. If my players are better and I have some skill then my play should work even if you know its coming...that's football.

[TombSong]

In single player games I can understand people wanting "adaptive AI" for the cpu to use.

However the human players team should never have "adaptive ai" The human team should have a offensive/defensive coordinator that alerts him to the opponents tendencies, then you call a play based on your "adaptive ai"(your brain).

I would hate to have called a play thinking my players are gonna do one thing, only for them to do something else based on what the cpu thinks is gonna happen.

[Exonerated]

Quote: Originally Posted by TombSong

In single player games I can understand people wanting "adaptive AI" for the cpu to use. However the human players team should never have "adaptive ai" The human team should have a offensive/defensive coordinator that alerts him to the opponents tendencies, then you call a play based on your "adaptive ai"(your brain). I would hate to have called a play thinking my players are gonna do one thing, only for them to do something else based on what the cpu thinks is gonna happen.

Only his PRC will go up if they see that specific route. And with higher PRC, players will read the play better. Thus, they will shut down plays that are repeated 5 times in a row.

This will make players like Asomugha god, even b4 adaptive ai.

[marktg30]

Quote: Originally Posted by Sven Draconian

Firstly, this better not be a factor if you run a play 3-4 or even 5 times. There really aren't that many "different" plays and having any sort of offense means you have a base package that you run.

Read this article and tell me that there aren't that many different plays.