Quote:
Originally Posted by Le Grande Orange
But with that AI following the same rules as does the human player? That's the key question. Often, a game will give the AI certain advantages that the human player doesn't get.
|
Yeah that's a very good point. I abhore games that "cheat" to make the game harder. Examples abound... Madden does it on All-Madden as admitted by EA. Tiger Woods rubberband opponent skills, et cetera. Many racing "sims" also change opponent drivers skills/behavior depending on the players performance.
In these cases the game requires some sort of physical dexterity on the player's part. That is often the justification for resorting to "cheating" to offer challenge to a wide-spectrum of players.
However, there is a case in point regarding Madden. Long-time players will tell you that for YEARS the play calling by the computer has been absolutely atrocious. It's not uncommon for the computer to call a run on 3rd and long or play 2 deep dime back zone on D on a 3rd and 1 -- not just sometimes but very often. Less cheating would be necessary if reasonably applicable plays were called in those situations.
But that's a different tangent.
No, I think it's quite possible to make a challenging game, especially a sports management SIM without the AI cheating. Let's just look at baseball GMing for example.
What does the AI have to do?
1. It has to evaluate potential trades decently.
2. It has to be able to create reasonable depth charts/batting orders.
3. It has to draft reasonably (using a lot of the same evaluation techniques applicable to trading).
4. It has to have some "knowledge" or "feel" for its organization's financial resources to determine how much $$$ to offer a FA or existing players renewing contracts without hurting its organization too much for the future.
5. It has to make decent in-game managerial decisions regarding pitcher replacement, pinch runners/hitters, defensive subs, et cetera.
If the AI does these 5 things reasonably well, I think it's de facto competative already. Do you agree with this premise?
Adjustments in "gamplay difficulty" are then already achievable depending on whether the user decides to manage the proverbial Yankees with nearly unlimited financial resources or the cash-strapped Royals.
Now, the only addition to these five areas, that I've posted about elsewhere, is mixing it up a bit. In other words, don't have 29 AI GMs all going after the same FAs and offering the same contract offers. Don't have them ALL wanting to draft the same players in a draft and value each position identically. Instead apply a GM "personality" filter that simply takes the default logic but then modifies it a bit so that for instance GM#1 prefers defense, GM#2 like big bats and prefers a strong SP#1 and 2 but then values its bullpen more than SP#3,4,5 and so on.
This variety makes the AI act more like real-life (so that it's not one human GM versus 29 clones) and also helps to cover up and/or alleviate issues like the good MRs remaining very deep into a draft (as discussed elsewhere.)
But anyway, it's just my $0.02. And I'm quickly learning these viewpoints are in the extreme minority around here, so I'm just about to stop posting anyway. I've only got 3 days left on my trial and it's looking grim for real MLB rosters which was the last thing I wanted to try with OOTP this year.
Good luck and take care.