whiners gonna whine
b1tchers gonna b1tch
part of any online game
Bernie, fix this crap
14 posts
• Page 2 of 2 • 1, 2
by bernieh » Wed Nov 09, 2011 3:55 pm
Hey guys, I hope I can shed some light on some things.
For what it's worth, I do think that the bullpen A.I. "isn't that bad". I'd even consider saying it's "pretty good", but I know anything I say in its defense only upsets its critics even more and doesn't change anyone's mind.
The only thing I'll say is that the settings DO work - just not all the time. And when they don't seem to work, there's usually SOME rationale - sound or not - behind it. I'm not going to try to defend the example you gave or any individual examples because sure, those could indeed be failures of the system. It's just that the A.I. is succeeding way more often than you may realize. We tend to notice more when it doesn't - understandably. Again, feel free to take my opinion for what it's worth or not worth.
The fact of the matter is, though, improving the A.I. is haaaaaaaaaard. Bullpen v2 took a [i:4216726604]year[/i:4216726604] of my life, and it was the most grueling work I've done in my career. I'm proud of what I accomplished (with the help of the focus group I worked with), but the constant, rigorous testing of all the changes I was making was enervating. It's never easy "just fixing" something with the computer manager because we always have to consider what repercussions any seemingly trivial change may have on the rest of the system. If the "Avoid" checkboxes removed players 100% of the time, you'd create immediate bench and bullpen shortages in the short term, and fatigue-related train wrecks in the long term. The computer has to (try to) be smart about when the checkboxes go into effect, and therein lies the rub. You could argue that it's not smart enough, and I wouldn't be able to prove otherwise. But it's just very hard to improve upon.
This all isn't to say that it will never be attempted again. As difficult as Bullpen v2 was, I still have plenty of motivation to work on bolstering the manager A.I. It's kind of like having kids - the passage of time blurs your memory of how awful babies are to deal with; otherwise you'd never have more. :) But any projects of this scale ultimately need to be approved by Strat-O-Matic, which needs to make sure things make the most sense for the future of the business. Who knows what will happen; they surprised me when they said they're backing the rewrite of the site. My time is booked on that for the next few months, though.
For what it's worth, I'm personally more concerned with the other complaints you've been sending in regarding the pinch-hitting logic. That part of the A.I. hasn't been looked at and the evidence I've been receiving from various people about its shortcomings are convincing. I'm with you, brother.
Clarifying some other points:
The answer to whether I have the authority to tinker with the computer manager A.I. is halfway in between what rburgh and AeroDave10 said. I do have that ability - but only with SOM's (strict) approval. I'd done plenty of things without their consent before and they were understandably not thrilled with some of it.
And rburgh, I disagree with your assessment of Bullpen v2 & 3. The interface of v2 was in the CD-ROM game before, but the overall manager A.I. was made [i:4216726604]much[/i:4216726604] smarter with the work that I did. SOM was initially uneasy that I was making it that much better than the CD-ROM game at the time. Regarding v3, though, I guess I'm not familiar with the advances the CD-ROM game has made since... please enlighten me. Thanks!
For what it's worth, I do think that the bullpen A.I. "isn't that bad". I'd even consider saying it's "pretty good", but I know anything I say in its defense only upsets its critics even more and doesn't change anyone's mind.
The only thing I'll say is that the settings DO work - just not all the time. And when they don't seem to work, there's usually SOME rationale - sound or not - behind it. I'm not going to try to defend the example you gave or any individual examples because sure, those could indeed be failures of the system. It's just that the A.I. is succeeding way more often than you may realize. We tend to notice more when it doesn't - understandably. Again, feel free to take my opinion for what it's worth or not worth.
The fact of the matter is, though, improving the A.I. is haaaaaaaaaard. Bullpen v2 took a [i:4216726604]year[/i:4216726604] of my life, and it was the most grueling work I've done in my career. I'm proud of what I accomplished (with the help of the focus group I worked with), but the constant, rigorous testing of all the changes I was making was enervating. It's never easy "just fixing" something with the computer manager because we always have to consider what repercussions any seemingly trivial change may have on the rest of the system. If the "Avoid" checkboxes removed players 100% of the time, you'd create immediate bench and bullpen shortages in the short term, and fatigue-related train wrecks in the long term. The computer has to (try to) be smart about when the checkboxes go into effect, and therein lies the rub. You could argue that it's not smart enough, and I wouldn't be able to prove otherwise. But it's just very hard to improve upon.
This all isn't to say that it will never be attempted again. As difficult as Bullpen v2 was, I still have plenty of motivation to work on bolstering the manager A.I. It's kind of like having kids - the passage of time blurs your memory of how awful babies are to deal with; otherwise you'd never have more. :) But any projects of this scale ultimately need to be approved by Strat-O-Matic, which needs to make sure things make the most sense for the future of the business. Who knows what will happen; they surprised me when they said they're backing the rewrite of the site. My time is booked on that for the next few months, though.
For what it's worth, I'm personally more concerned with the other complaints you've been sending in regarding the pinch-hitting logic. That part of the A.I. hasn't been looked at and the evidence I've been receiving from various people about its shortcomings are convincing. I'm with you, brother.
Clarifying some other points:
The answer to whether I have the authority to tinker with the computer manager A.I. is halfway in between what rburgh and AeroDave10 said. I do have that ability - but only with SOM's (strict) approval. I'd done plenty of things without their consent before and they were understandably not thrilled with some of it.
And rburgh, I disagree with your assessment of Bullpen v2 & 3. The interface of v2 was in the CD-ROM game before, but the overall manager A.I. was made [i:4216726604]much[/i:4216726604] smarter with the work that I did. SOM was initially uneasy that I was making it that much better than the CD-ROM game at the time. Regarding v3, though, I guess I'm not familiar with the advances the CD-ROM game has made since... please enlighten me. Thanks!
- bernieh
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
by nevdully's » Wed Nov 09, 2011 6:14 pm
Bernie,
Just a lessening of pitcher fatigue back to what it was (singles) would keep the routine scoring of 15+ runs a game down. This would keep hitting and pitching stats more balanced while probably not effecting w-l outcomes very much at all....I could go on....and on, and in better detail but I'll leave that for lunch. ;)
Thanks for all you've done so far.
Just a lessening of pitcher fatigue back to what it was (singles) would keep the routine scoring of 15+ runs a game down. This would keep hitting and pitching stats more balanced while probably not effecting w-l outcomes very much at all....I could go on....and on, and in better detail but I'll leave that for lunch. ;)
Thanks for all you've done so far.
- nevdully's
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
14 posts
• Page 2 of 2 • 1, 2
Return to Strat-O-Matic Baseball: All-Time Greats
Who is online
Users browsing this forum: No registered users and 20 guests