|
||||
|
|
OOTP 25 - General Discussions Everything about the brand new 25th Anniversary Edition of Out of the Park Baseball - officially licensed by MLB, the MLBPA, KBO and the Baseball Hall of Fame. |
![]() |
|
Thread Tools |
![]() |
#21 |
Hall Of Famer
Join Date: May 2016
Location: St Petersburg Florida USA
Posts: 6,657
Infractions: 0/2 (3)
|
Do some teams have computer GMs that are stronger or weaker than others? If yes, has it been considered giving the game a switch to allow stronger or weaker competition from the computer GMs?
__________________
. Pirates Play Moneyball 1951-2008 Ratings and League Totals Modifiers A new pre-calc file! (Applause!) Settings and Auto-calc Popular Opinions and Undisputable Facts The cover up is usually more damaging than the original act. |
![]() |
![]() |
![]() |
#22 | |
Hall Of Famer
Join Date: May 2004
Posts: 10,607
|
Again, my biggest issue, aside from the standard moral issues with AI, which I’m not sure completely apply here, is that machine learning is going to take all the variables and stick it into a big black box and come out with something… that will surely be better at winning OOTP but won’t necessarily feel like a good baseball opponent. AI doesn’t “care” about what looks good or bad in a baseball sense. If it finds that switching the DH into the field every game and batting the pitcher from the 7th on is a winning strategy, it will use it. If it discovers that you should bat your worst hitter 2nd, it’ll do that. Actual managers have to work to not get fired; all an AI will try to do is win.
And I get that people would find that interesting but I think at the end of the day what you’d find is a way to exploit OOTP itself much more than baseball. Which, hey, I guess, if an AI uncovers some massive exploit (let’s say it decides to take on every pitch) that the devs can fix, that could be a good thing (although in my example here the “fix” has long been “just don’t do this”). On the other hand, pushing the devs to fix stuff that’s not exactly broken until a ML model exploits it might be a waste of time, and an AI that does weird and screwy and non baseball things is going to get complained about a *lot* even if you can demonstrate its actions win more games. Anyway, aside from physical constraints there’s no necessarily anything stopping someone with a copy of the game to fire up Amazon’s ML system and put together a system to run a hundred seasons of AIs competing against each other. No, you won’t have the reams of other people’s data to work with but you’ll have your own.
__________________
Quote:
|
|
![]() |
![]() |
![]() |
#23 |
Hall Of Famer
Join Date: May 2016
Location: St Petersburg Florida USA
Posts: 6,657
Infractions: 0/2 (3)
|
Good post. And I don't hate to agree with Syd! LOL.
Anyway, a computer manage like you describe would be hated by some people more than they hate sabermetrics!
__________________
. Pirates Play Moneyball 1951-2008 Ratings and League Totals Modifiers A new pre-calc file! (Applause!) Settings and Auto-calc Popular Opinions and Undisputable Facts The cover up is usually more damaging than the original act. |
![]() |
![]() |
![]() |
Bookmarks |
|
|