An experiment on CPU offensive AI and sliders.....

[ParisB]

Re: An experiment on CPU offensive AI and sliders.....

Well I know what you're saying, that you should still be able to see instances of being able to blow down the opponent with overpowering stuff...

But I think the bigger issue is the level of difficulty. I feel that I can bring nasty stuff over the middle and see the CPU whiff on All-Star (occasionally and probably at a good realistic rate), but it seems (from your posts) that it doesn't happen on the higher difficulty levels.

I'm sure there's a fine line somewhere where they have to make the CPU tough and perhaps one of those methods is to punish anything over the plate more often.

I don't know....

[Phoenixmgs]

Re: An experiment on CPU offensive AI and sliders.....

I'm sure there's a fine line somewhere where they have to make the CPU tough and perhaps one of those methods is to punish anything over the plate more often.

THAT exactly. Meter pitching is too accurate so the CPU naturally has to be good a hitting pitches that get a good chunk of the plate no matter who the pitcher is.

[NEOPARADIGM]

Re: An experiment on CPU offensive AI and sliders.....

My last 10 playing on default All-Star:

W 6-3
W 8-4
L 4-5
L 5-7
W 3-2
W 4-3
W 17-8
W 7-1
L 11-5
L 11-0

Just throwin' that out there.

I'm the last person to tell anyone how to play, or what to do more generally, but maybe if you'd just lighten up a little, play the game for what it is, threads like this wouldn't get started in the first place, and you'd be off playing, and having fun with results like those above, instead of analyzing every minute detail and being Dr. Science or whatever this thread is about.

My .02

[swaldo]

Re: An experiment on CPU offensive AI and sliders.....

Man I lost track of this thread because I didn't know it was moved to the slider section. Anyway, it's great to hear comments from Brian, it literally made me run and get a cup of coffee because I knew it was going to be a good read. This is the first time I can remember a dev addressing this - other than a denial the issues we think we're seeing don't exist. This is a positive step and from a dev's perspective I'd be curious to know the following:

1. What makes a guy like Philly want to save a partially completed game, then play it over & over 30 times? And what's with all these tests?

2. Why do people spend hours to capture & upload video or take screen shots and post their complaints hoping someone out there is listening?

3. In '09 why were there were multiple, heated 100+ page threads regarding "comeback code?"

4. Why do so many people continue to get posts deleted or outright banned for stating their feelings about this sort of issue? And if there was no censorship what would people be saying?

5. Why do several respected members of OS (who are sports and baseball nuts) prefer to play 2K or a Japanese baseball game with no verbal or written english in it?

6. Why do guys like me put a computer on my lap while playing The Show and log stats in excel? And why after 260 innings of data was my pitchers ERA 3.00 when my team was losing, but 6.58 when my team was winning in '09? Ok so people get complacent when they are winning which could allow the CPU to perform better. However, I find it's actually the opposite way around. When I'm winning I have to focus and concentrate because I always have this feeling the CPU is going to get me. However, if I'm losing I can sit back, relax and not worry about what I'm throwing because the CPU bats always seem to get cold. Make a mistake and throw a meatball? No problem, it will be popped up.

Currently I'm running a new "test" where I changed batters to 99 ratings across the board on one team, and the other team has all zero ratings. Currently, the 99 rated batters are as effective as a pitcher at the plate. Is this because the zero team is doing nothing so the CPU wants to keep things close? The sample size is low but thus far the AI acts similiar to the legend max tests Philly & myself conducted. Really though, and I think I speak for everyone - we'd rather just be having fun playing the game than running tests.

I don't expect answers to these questions. I only mention them because it seems obvious that people are having problems, and if the game is programmed with truly organic AI then we need help in understanding the game in order to play it properly. It's refreshing to hear Brian's comments and more behind the scenes information, tips and open dialog would be appreciated. I read, absorbed and pondered everything Brian said and overall I think this game is brilliant.

Finally, the whole reason I revisited this thread...

I stumbled across the post below on another site from a user who is not a troll or basher. He loves the game, has played it a ton and had a goal to get a no hit trophy. In other words, when he was pitching perfectly he absolutely tried his best to get the CPU out. I think his experience pretty much sums up how alot of people feel when playing this game...

"I played 2 complete seasons in last years game. Both times I won about 100 games. So I never cared if I lost one out of every 5 games in some BS 8th innning rally. However, I flirted with the Perfect Game trophy 15 to 20 times until I finally got it. After nearing perfection 15 times or so in the 7th inning, I picked up on how to get the trophy.

This is how I got the trophy:

I would play normally until I was perfect in the 5th inning. I would check who was pitching and his stamina. I would check my current winning streak, and what he did his last 3 outtings. If I was on a 5 game winning streak, or my pitcher was 3 and 0 this month (These are the things the game will point out for you normally) I would say forget it, not gunna happen. And just play the game.

If I just lost the game before this, and my pitcher had a bad game last time then I would look at the score of the current game. If I was up by 4 or 5 runs, I would say forget it and keep playing. If I only had a 1 or 2 run lead I would say to myself, ok, time to get this trophy.

For the rest of the game I would bunt with everyone who came up to the plate, and put in my fastest players in every defensive position. Fielding doesn't seem to effect who makes errors and who doesnt, that was just my observations. This game will make Jeter make an error if you're up by 5 or 6 runs in the late innings. So it's all about getting to the ball the fastest way possible.

And after bunting for 3 innings I finally got the perfect game trophy. And I did it with Freddy Garcia, not even close to the ace of any staff by last years stats. And he was out of energy by the 8th inning, probably looking at like 100+ pitches.

So take it for what its worth. I had many a game where I was perfect for like 7 innings but I was up by 6 or 7 runs. And they would always get to the pitcher late. And yes, that happens all the time in real baseball, and it makes perfect sense. So take it for what it is. I got my perfect game trophy in a game where I had 1 run on 4 hits. And bunted for 4 innings.

And most of my shutouts and complete games that I would get in the season were always pitching duels. Where the script would constantly remind you of how both starters showed up to play today, it's a real duel, anything you can do I can do better, we are in for a real treat today, what a match up this is, it's like they are facing off against each other. All those comments are all you hear in most Shutouts and Complete games.

But hey, maybe someone got their trophy in a 15 to 0 game."

[Phoenixmgs]

Re: An experiment on CPU offensive AI and sliders.....

^^^
Yeah, a mod moved it to the slider subforum for no discernable reason. I agree the game is scripted to a degree. I haven't played enough CPU games this year to be able to prove anything, but the first 9 games of my season were ridiculous and now offense has dramatically dropped off since then for both the CPU and I, it just doesn't seem right the way some things happen in the game, you know what I'm talking about. I picked up Resonance of Fate and I've been playing it non-stop. I know last year certain games were scripted to be low scoring or high scoring. I would purposefully replay certain games in 09 that felt conveniently offensive or pitcher friendly. And, every time I replayed it, the same kind of game happened, slugfest or pitching duel, and it felt very artificial. One season, my divisional series against the Dodgers was a pitching duel every game and every run was huge. Then, in the league championship series against the Cardinals, every game was a complete slugfest, like 16-14 type slugfests. Every game of the series the CPU could not get me out and I couldn't get the CPU out. Pretty much whoever had last at-bats won the game. I won the series by scoring 7 runs in the bottom of the 9th.

Currently I'm running a new test where I changed batters to 99 ratings across the board on one team, and the other team has all zero ratings. Currently, the 99 rated batters are as effective as a pitcher at the plate. Is this because the zero team is doing nothing so the CPU wants to keep things close? The sample size is low but thus far the AI acts similiar to the legend max tests Philly & myself conducted.

That's real disappointing to hear. I don't think I'm really going to play this year's game much. There was so much game stacking early this year that I have plenty of games to get to (Heavy Rain, God of War 3, Final Fantasy XIII, etc.). My league is about to start soon, so I'll be spending time playing the game online where the AI issues are no longer there. We waited for the online issues to be resolved. I'll try to get some RTTS in at some point.

[swaldo]

Re: An experiment on CPU offensive AI and sliders.....

And Phoenix I agree with you about the CPU not chasing enough. Especially on 0-2 or 1-2 counts batters have to protect the plate but I don't see this enough in the game. Too often I'll shave the corners with great pitches (just missing a strike) but somehow end up in 3-2 counts.

In real life Rex Hudler announces Angel games and when batters get caught looking on questionable strikes he gives them hell. "You can't argue that, you gotta be hacking on 0-2!" Batters might take a chance on a borderline pitch on a 2-2 count, but a guy has to have great vision to watch them go by on 0-2 or 1-2.

[Chef Matt]

Re: An experiment on CPU offensive AI and sliders.....

You know guys, in all honesty I'll say I saw the scripted gameplay/CPUCB code/rubberband AI, whatever you want to call it, in last years game. It was there IMO and the game to me was unplayable becuase of it. I traded it in within a month of having the game.

But this year I truly don't see it.

I play on all default sliders HOF. I just don't see anything like '09. I've had 3 save opps in my season so far and all of them were a breeze to get through. I never felt threatened, never see that dreaded 9th inning "error" to put a runner in scoring position like last year etc..

I just recently beat the Rangers in my season 9-2, there was no way I'd ever get a score like that in '09. Once I got up by about 5 I just started pitching for contact and there was never a single threat with the exception of a Solo 500 ft shot by Vlad on a pitch that I forgot to set the location on. (I deserved that one)

When I first got this game this year it took me about 10 games to lose the feeling that this game was scripted, and considering how gun-shy I was that's a feat. I was WAITING for this game to do it again but I can say now that crazy, crappy feeling of "no matter what lead I get they're going to comeback" is gone.

Thank goodness too because this is the BEST BB game I personally have ever played.

Im just tryin to help, I know how hard it is to get that feeling out of your head, but for me it's gone.

I'm not at all trying to be fecicous, but maybe your just reading too much into it? IDK.

If you guys have any questions on my style of play or what my game settings are PM me. I'm no baseball guru I just like the game but I'm willing to help people lose that edgieness when it comes to this game because it is incredibly more enjoyable when the only thing you get frustrated at are swinging at ball four in the dirt.
JM .02

Cheers

[phillyfan23]

Re: An experiment on CPU offensive AI and sliders.....

this COULD only exist in Legend mode who knows....ever since I switched to HOF pitching , I have been blown out and blown out the cpu themselves.

On legend, When i struggled hitting, so did the cpu and when I raked so did the cpu.

Not so with HOF pitching level. I still hit on Legend, but the results so far have been great.

Lost 9-1 8-3, and won games 7-0, 8-2.

On legend pitching , where i did the initial testing, games were always close. and the cpu went numb whenever I couldn't muster any hits.

[swaldo]

Re: An experiment on CPU offensive AI and sliders.....

I saw the scripted gameplay/CPUCB code/rubberband AI, whatever you want to call it, in last years game. It was there IMO and the game to me was unplayable becuase of it. I traded it in within a month of having the game.

But this year I truly don't see it.

Glad to hear you're having good results. I'm not ready to say the same thing myself. Below is a log from a game where the players were edited and mixed with half 99 rated batters and half zero rated batters across the board. Why this is so I'll save for another day.

This was a CPU vs CPU game and it was a low scoring game. In the bottom of the 9th the Dodgers needed two runs to win and got it. Ausmus (who started out the 9th with a HR) is a 99. Loney, Ethier, Belliard and Kemp are all zeros and the bunts they laid down were perfectly placed...

Text-April4.jpg

I think it's great that anything can happen in this game. However, wonky innings like this seem to happen too often & tend to benefit the losing team more often than not.

When I pitch against the zero batters they can barely get it out of the infield. You won't see the laser darts, instead they hit balls softly and the only hits are seeing eye singles or bloopers. Their batting average is less than .100 against my pitching and they haven't been able to scratch a run over unless a 99 also bats in the inning. In the CPU vs CPU games the zeros scored 2 of 37 total runs.

I'm not saying this proves anything, it's just an example of why I scratch my head at some of the results in this game.