More AI oddness

[Malleus Dei]

Originally Posted by Markus Heinsohn

Having a bad day, eh?

Any day where I see a bug that was supposed to be gone return is a bad day.

I do not argue with you that OOTP is very good. But I am also German by ancestry and heritage, just not by birth, and as such I expect everything to be far better than just very good.

You should be thankful. Perfectionist customers make for better products.

[johnnybourbon]

i've bit my tongue on this one for a while. I kinda wanna stand up for Markus, as I'm a programmer myself. This particular bug irks me. Obviously, in just the trade screen, it's nearly impossible to code for every possible contingency of player combinations and ai needs.

The issue I have is that from (in my own experience as well) the programmer down to the tester, you can't anticipate a user trying to do something that to the user, is something they do all the time, but isn't really something that should occur in the program in the first place.

This is something that I get riled up about because users of MY program say things like "you never fix this", "you never test this", "it NEVER works", etc...when their particular bug only happens to 1-3% of the users because they do things their way every time, which isn't necessarily the right way.

I see the same reactions here. For this bug...I can't help but say, "sure, the AI shouldn't have made that trade in the first place, but, seriously, who trades players that even a less-than-perfect AI will surely not want at even their highest minor league?" That being said, a lot of us here have learned how to fleece the AI in the trade screen. The fact that the AI is reacting badly to a loophole that the user is exploiting should hardly have the finger pointed at Markus.

[jbergey22]

My only reply to that because you make a lot of very good points.

Cant Markus code a value on a player?

On a scale of 1-100 lets just say a 5 star prospect has a value of 100 while a 1 star 26 year old Triple A pitcher has a value of 1 or something like that therefore he is unimportant in the trade either way.

If the computer is taking less than what they are giving up they dont accept the trade.

I feel very confident that he could come up with a great numbering system that puts a value on players actual worth so we dont see this type of stuff.

[johnnybourbon]

Originally Posted by jbergey22

My only reply to that because you make a lot of very good points.

Cant Markus code a value on a player?

On a scale of 1-100 lets just say a 5 star prospect has a value of 100 while a 1 star 26 year old Triple A pitcher has a value of 1 or something like that therefore he is unimportant in the trade either way.

If the computer is taking less than what they are giving up they dont accept the trade.

I feel very confident that he could come up with a great numbering system that puts a value on players actual worth so we dont see this type of stuff.

But what you're proposing is what I've found to be the loophole (granted, not with the trade difficulty through the roof). The idea is by a value system, if you want a 4 star player, you'd hafta give up 4 1s, which in itself is illogical and logical at the same time.

[jbergey22]

Originally Posted by johnnybourbon

But what you're proposing is what I've found to be the loophole (granted, not with the trade difficulty through the roof). The idea is by a value system, if you want a 4 star player, you'd hafta give up 4 1s, which in itself is illogical and logical at the same time.

Well I would value change my values so that 1s are basically crap so ud have to give up 100 1's actually in my example which Im sure Markus would put a limit on how many players can be traded at one time.

So 5 20s for 1 100 doesnt seem terrible to me if the value of a 20 actually means something(mid tier prospect or MLB starter with a decent contract)

100 5 star prospect or top tier arbitration year MLB player
70 4 star prospect
40 3 star prospect
10 2 star prospect
1 1 star prospect or MLB fillers
0 MLB players with BAD contracts

Use that as a platform and go from there

[Flying Dutchmen]

Originally Posted by jbergey22

Well I would value change my values so that 1s are basically crap so ud have to give up 100 1's actually in my example which Im sure Markus would put a limit on how many players can be traded at one time.

So 5 20s for 1 100 doesnt seem terrible to me if the value of a 20 actually means something(mid tier prospect or MLB starter with a decent contract)

100 5 star prospect or top tier arbitration year MLB player
70 4 star prospect
40 3 star prospect
10 2 star prospect
1 1 star prospect or MLB fillers
0 MLB players with BAD contracts

Use that as a platform and go from there

The point is that someone would, and trust me they would, find a way to game that scoring system as well. Human players (thousands of them with differing mindsets and strategies) will probably always be able to outthink a single AI design. And that's without the following: what about trading between teams with different budgets? Involving a team one piece away from making a championship run? Teams with no minor leagues? Historically evolving and increasing budgets? No 40-man roster? No option years? A salary cap? Differing salary structures?

With the number of options that OOTP has, I can easily see how it's hard to come up with an AI that will work with every conceivable combination of options that people will come up with. Clearly it needs to, because if it doesn't, those people will bombard the forums with complaints about the "woeful" AI that "must" be fixed.

[PSUColonel]

Originally Posted by Markus Heinsohn

Having a bad day, eh? Come on, issues like this are very rare in OOTP... and compared to other sports sims its AI is extremely good, especially considering it has to handle all the different rules and options.

I agree with this too!!!