Ask the devs a question

[Bornindamecca]

Re: Ask the devs a question

Hey Mike,

How do you feel about the "Robo-AI" right now? In football games it's black and white with the QB: In real life if you rush the QB, you'll often force him into erratic throws, and if you get constant pressure on him, he'll be more mistake prone until something changes to allow his confidence or rhythm to return.

In basketball it's not quite as black and white, but right now I think we're still suffering from the same situations in basketball games. It seems like to make the AI difficult, player awareness gets raised to a level that takes it outside of sim. For example, even on the hardest level of a basketball game, Corey Maggette is going to commit offensive fouls if it's a sim.

Is there a way we can force AI players to make mistakes and bad decisions but stay within the logic of fundamentals? We touched on this in the late game AI topic. We all know the basic tenets of what a team should do if they are down or up by 3, but what about when we're playing a lesser team? What if a starter is injured or out, and a scrub is forced to execute in a tight situation? Is Martell Webster going to execute his late game defensive assignment as Brandon Roy? Will inexperienced bigs make silly fouls? Young PGs make bad passes under pressure?

I'm hoping for a more organic experience, where difficulty still includes some "humanity" in the AI, and the reward for great defense is the collapse of the opponent's execution, and perhaps some desperation and heroics on the part of superstars with those tendencies. Throw in some frustrated coach animations, and we have a fresh experience.


But I don't know how tough this is from a programming perspective, which is why I ask you.

[pbz06]

Re: Ask the devs a question

Hey Mike,

How do you feel about the "Robo-AI" right now? In football games it's black and white with the QB: In real life if you rush the QB, you'll often force him into erratic throws, and if you get constant pressure on him, he'll be more mistake prone until something changes to allow his confidence or rhythm to return.

In basketball it's not quite as black and white, but right now I think we're still suffering from the same situations in basketball games. It seems like to make the AI difficult, player awareness gets raised to a level that takes it outside of sim. For example, even on the hardest level of a basketball game, Corey Maggette is going to commit offensive fouls if it's a sim.

Is there a way we can force AI players to make mistakes and bad decisions but stay within the logic of fundamentals? We touched on this in the late game AI topic. We all know the basic tenets of what a team should do if they are down or up by 3, but what about when we're playing a lesser team? What if a starter is injured or out, and a scrub is forced to execute in a tight situation? Is Martell Webster going to execute his late game defensive assignment as Brandon Roy? Will inexperienced bigs make silly fouls? Young PGs make bad passes under pressure?

I'm hoping for a more organic experience, where difficulty still includes some "humanity" in the AI, and the reward for great defense is the collapse of the opponent's execution, and perhaps some desperation and heroics on the part of superstars with those tendencies. Throw in some frustrated coach animations, and we have a fresh experience.


But I don't know how tough this is from a programming perspective, which is why I ask you.

That's almost exactly one issue I mentioned in my post, just worded differently. I feel like I'm playing the exact same AI or CPU every game, regardless if I'm the Celtics/Lakers or Bobcats/Thunder. If I'm playing on Superstar, I'm simply at the mercy of the CPU's programmed shooting percentages no matter who I'm going up against. It won't be any easier or harder regardless of what team or player I'm cafing against. A Felton or Matt Caroll are just as likely to scorch me as a Garnett or Kobe, just because the difficulty level is programmed to make it hard for me. Even on lower difficulty levels, it just feels like the same opponent but with a different skin texture.

In real life, teams play differently and at different paces. The Spurs always slow it down and go through Duncan. The Warriors play recklessly etc. Teams also differ in clutch situations and make more or less mistakes than others.

[Beluba]

Re: Ask the devs a question

Hey Mike,

How do you feel about the "Robo-AI" right now? In football games it's black and white with the QB: In real life if you rush the QB, you'll often force him into erratic throws, and if you get constant pressure on him, he'll be more mistake prone until something changes to allow his confidence or rhythm to return.

In basketball it's not quite as black and white, but right now I think we're still suffering from the same situations in basketball games. It seems like to make the AI difficult, player awareness gets raised to a level that takes it outside of sim. For example, even on the hardest level of a basketball game, Corey Maggette is going to commit offensive fouls if it's a sim.

Is there a way we can force AI players to make mistakes and bad decisions but stay within the logic of fundamentals? We touched on this in the late game AI topic. We all know the basic tenets of what a team should do if they are down or up by 3, but what about when we're playing a lesser team? What if a starter is injured or out, and a scrub is forced to execute in a tight situation? Is Martell Webster going to execute his late game defensive assignment as Brandon Roy? Will inexperienced bigs make silly fouls? Young PGs make bad passes under pressure?

I'm hoping for a more organic experience, where difficulty still includes some "humanity" in the AI, and the reward for great defense is the collapse of the opponent's execution, and perhaps some desperation and heroics on the part of superstars with those tendencies. Throw in some frustrated coach animations, and we have a fresh experience.


But I don't know how tough this is from a programming perspective, which is why I ask you.

Great post, Born. This is exactly what we're aiming for with our AI... especially with respect to difficulty level and player differentiation. The problem is, everyone (the comp included) has been trying so hard just to build "good basketball AI." And up to this point, no one's really invested in building intentional and believable "bad AI." So when it comes time to tune difficulty levels, developers have to cheat the systems so that the effect comes across in some form or fashion. But that just yields the problems you're mentioning -- players who react unrealistically fast, can do superhuman things, or on the other hand, are nerfed to the point of futility. I wouldn't say what you're looking for is necessarily hard to do... but it does require investment and has to be accounted for from day 1 when you get into feature development. Having said all that, we are striving for that level of depth in Live '10 and Dynamic DNA will be the primary tool we use to accomplish it.

[Bornindamecca]

Re: Ask the devs a question

Thanks for the answer, Mike. I look forward to you guys putting the work in and getting results. I'm not a game designer, so I humbly leave that to the experts, but I am a storyteller so I wanted to share a thought on that.

We touched on it a bit during the late-game-AI discussion, but I wanted to expand on the point. You phrased it well when you said "believable AI", which is really what we're all talking about. I think taking the simulated human error deeper than the surface is the key. When the AI makes poor decisions(which, if you're emulating Jared Jeffries, you're really talking about great AI, haha), the user will buy into it if the action is reflected in the crowds, the coach animations, the player animations and the commentary. A "my bad" animation on a stupid foul, or a good long Popp-esque stare down from the coach would clue in a majority of the gamers. Commentary example: "They're trying to go one on one right now, but that's not how they established their lead in the first half." or "That was a silly turnover; this team is overdribbling.."etc etc

Things like this retain suspension of disbelief in the virtual NBA world. On the super AI side of things, I think players are willing to accept that as long as its stated. When the levels are called All Star, Superstar, etc...people are looking at one kind of difficulty increase, and perhaps rightfully so. But when you call something "god mode" or "Impossible", people understand that it's not really about you beating fair AI, but this is for those select few who have mastered the best simulation you guys can offer, and need a perfectly executing AI for a real challenge.

We all know that some people will miss out on what you're doing, but I'm sure you don't take on these deadlines and challenges with the expectation of 100% approval. If this realistic, organic, human-emulating AI is expressed throughout every facet of the presentation, it'll be a successful step forward in Live, but also in sports gaming overall. DNA is certainly a great, great place to start.

Thanks again for taking the time, man.

[Beluba]

Re: Ask the devs a question

Thanks for the answer, Mike. I look forward to you guys putting the work in and getting results. I'm not a game designer, so I humbly leave that to the experts, but I am a storyteller so I wanted to share a thought on that.

We touched on it a bit during the late-game-AI discussion, but I wanted to expand on the point. You phrased it well when you said "believable AI", which is really what we're all talking about. I think taking the simulated human error deeper than the surface is the key. When the AI makes poor decisions(which, if you're emulating Jared Jeffries, you're really talking about great AI, haha), the user will buy into it if the action is reflected in the crowds, the coach animations, the player animations and the commentary. A "my bad" animation on a stupid foul, or a good long Popp-esque stare down from the coach would clue in a majority of the gamers. Commentary example: "They're trying to go one on one right now, but that's not how they established their lead in the first half." or "That was a silly turnover; this team is overdribbling.."etc etc

Things like this retain suspension of disbelief in the virtual NBA world. On the super AI side of things, I think players are willing to accept that as long as its stated. When the levels are called All Star, Superstar, etc...people are looking at one kind of difficulty increase, and perhaps rightfully so. But when you call something "god mode" or "Impossible", people understand that it's not really about you beating fair AI, but this is for those select few who have mastered the best simulation you guys can offer, and need a perfectly executing AI for a real challenge.

We all know that some people will miss out on what you're doing, but I'm sure you don't take on these deadlines and challenges with the expectation of 100% approval. If this realistic, organic, human-emulating AI is expressed throughout every facet of the presentation, it'll be a successful step forward in Live, but also in sports gaming overall. DNA is certainly a great, great place to start.

Thanks again for taking the time, man.

that's an excellent point, and something we've identified as an area our dev team needs to improve. Sean, our line producer, has been emphasizing the need for the different game areas (especially gameplay & presentation) to do a better job at supporting each others' work and not operating in silos. so we're definitely expecting the package as a whole to be much more cohesive than it has been in the past... and to your point, take the next step toward a more organic experience.

[Sovartus]

Re: Ask the devs a question

Beluba,
In my opinion, one of the things that has plagued bball games in the past has been inability to overcome cheap tactics with strategy and hoops knowledge. I understand that you have to cater to the audience and a majority of players only watch the game on TV but it becomes very frustrating when a person knows bball and gets blown out by someone who simply exploits the programming flaws to gain an advantage. I realize that no game will ever be perfect and arcade style players will "find a way" but is there a happy medium? I have never been a fan of lobby ball and mainly stick with people I know but it would be nice to have an option to stop these "cheap" tactics that people use to win. It's great that post defense exists in Live but how do we make it harder for people to just beat you by getting to the rack almost every time. NBA statistics show that most points (normally around 70%) are scored outside of the paint but 90% of shooting fouls happen in the paint. I'm sure that is tough to leverage in a game since the player is ultimately in control of where shots are taken. I just wonder if there is something that a defender can do or be put in place in Live 10 to force more jumpers? (Maybe at least have it as a settings option?)

I'm very excited about DNA because there are so many possibilities and it seems like you are rewarded for playing the way the players do in real life. I would hope that this has more impact in the future but for now I would like to get your feedback on this issue if you have time.

[Beluba]

Re: Ask the devs a question

Beluba,
In my opinion, one of the things that has plagued bball games in the past has been inability to overcome cheap tactics with strategy and hoops knowledge. I understand that you have to cater to the audience and a majority of players only watch the game on TV but it becomes very frustrating when a person knows bball and gets blown out by someone who simply exploits the programming flaws to gain an advantage. I realize that no game will ever be perfect and arcade style players will "find a way" but is there a happy medium? I have never been a fan of lobby ball and mainly stick with people I know but it would be nice to have an option to stop these "cheap" tactics that people use to win. It's great that post defense exists in Live but how do we make it harder for people to just beat you by getting to the rack almost every time. NBA statistics show that most points (normally around 70%) are scored outside of the paint but 90% of shooting fouls happen in the paint. I'm sure that is tough to leverage in a game since the player is ultimately in control of where shots are taken. I just wonder if there is something that a defender can do or be put in place in Live 10 to force more jumpers? (Maybe at least have it as a settings option?)

I'm very excited about DNA because there are so many possibilities and it seems like you are rewarded for playing the way the players do in real life. I would hope that this has more impact in the future but for now I would like to get your feedback on this issue if you have time.

You're exactly right, sovartus. Online play too often comes down to who can exploit the cheese in the game most effectively. We do our best to build cheese-free games, but obviously haven't been successful yet. The good news is that we've laid out a plan to combat the typical PIP issues and have thought of some good ways to reward players who play good fundamental basketball. And yes, DNA will play a large role in that plan. I'm not ready to guarantee that Live '10 will be a perfect, exploit-free experience, but I think we're going to get very close.

[Court_vision]

Re: Ask the devs a question

Great thread and great posts from Born and Sov.

Really interesting to read Beluba's responses. I know nothing about computers or programming, so it's kind of cool to hear what goes into the making of one of thse games

One question Mike...how hard is genuine guard help D to implement in a basketball game?

LIVE has it done pretty well at the moment - 2K9 basically doesn't have it at all.

Is it really hard to implement? Or is it a design choice not to include it? (i.e. making guards stay on shooters cuts down on open 3's online etc)

[RAZRr1275]

Re: Ask the devs a question

Hey Mike, which things are you looking to implement in the future (not necessarily 10 but possibly after) and which out of those do you think will be the hardest and why?

[Beluba]

Re: Ask the devs a question

Great thread and great posts from Born and Sov.

Really interesting to read Beluba's responses. I know nothing about computers or programming, so it's kind of cool to hear what goes into the making of one of thse games

One question Mike...how hard is genuine guard help D to implement in a basketball game?

LIVE has it done pretty well at the moment - 2K9 basically doesn't have it at all.

Is it really hard to implement? Or is it a design choice not to include it? (i.e. making guards stay on shooters cuts down on open 3's online etc)

Well, it would never be a design choice NOT to include proper help D and rotation, but yeah, those aren't the easiest things to implement. Thankfully, we have some very talented AI engineers who really understand defensive concepts. There are a few tweaks we need to make, like making sure weakside defenders don't sag too much on great 3PT shooters... but I've always appreciated this aspect of Live, even before I got here.

[Beluba]

Re: Ask the devs a question

Hey Mike, which things are you looking to implement in the future (not necessarily 10 but possibly after) and which out of those do you think will be the hardest and why?

oh man. I wish I could answer this cause there are a ton of exciting ideas we've tossed around that don't fall into this year's scope... or even next year's. But for obvious reasons, I gotta keep 'em in the vault. Two things I really love about EA are that we're really good at creating long term plans and we're always on the cutting edge with regard to technology. So I've really been encouraged to dream big, which is very refreshing for a game designer. Those are just 2 of the many benefits of working at a huge company with a lot of shared resources.

The hardest thing to implement? Well, that would probably be my Star Wars-like basketball court hologram interface. That one might take a while.

[RAZRr1275]

Re: Ask the devs a question

oh man. I wish I could answer this cause there are a ton of exciting ideas we've tossed around that don't fall into this year's scope... or even next year's. But for obvious reasons, I gotta keep 'em in the vault. Two things I really love about EA are that we're really good at creating long term plans and we're always on the cutting edge with regard to technology. So I've really been encouraged to dream big, which is very refreshing for a game designer. Those are just 2 of the many benefits of working at a huge company with a lot of shared resources.

The hardest thing to implement? Well, that would probably be my Star Wars-like basketball court hologram interface. That one might take a while.

I thought that you couldn't say anything but I tried it just in case. I still got some good info out of this post though. It's really good that you guys know where your going in the future and know what you have to improve on. I'm really excited about this series. And the best part is with DNA I won't have to do any rosters at all


Now that would be awesome. Ship a virtual reality suit with the game and play on a virtual reality court. That would just be insanely cool. To bad it probably won't be until 2020-30 or something

[RAZRr1275]

Re: Ask the devs a question

Also how do you feel about DNA? What do you feel are it's advantages or limits?

[Mintsa]

Re: Ask the devs a question

Beluba....

Will you guys be or have talked about adding in stuff like tip ins & put back dunks for Live 10 ???

They really add to the gameplay. It takes the rebounding game to a whole different level.

Also, what about stuff like jump balls after a defender is able to reach in and get a hold of the ball from the offensive player ???

And animations like saving the ball (diving) and tiping it to a teammate ??

These are all just little things, but add a ALOT to make you feel like your really playing in the game. Remember, all these "little" animations can make a HUGE difference.

[bls]

Re: Ask the devs a question

What in the world kind of bug is in this game??????? Every game I'm playing there leaving a guy wide open. 2 guys guarding the same guy???? What is up with this? I just don't understand, why this problem is occurring now. Anybody else experiencing this?