Biased AI?

[Voriax]

I created a tiny scenario to test something that i've pondered now and then. I made a map that had one victory flag surrounded by trees and in the same hex with the flag is an entrenhced German AT-rifle unit. On the opposite side there was a Soviet armoured car. HQ units and other units in the atr/BA-20 platoon were immobilized at the ends of the map so they didn't interfere.

I ran the game 10 times as Germans, and every time the Russian vehicle arrived at the precisely same hex, and _every_ time the ATR team assaulted with 79% hit probability.

Then I switched to russian side, and ran that car 10 times to the same hex, and the ATR team assaulted 6 times with 24% prob. and 4 times with 12% prob. Now as I didn't change any settings between games, shouldn't the to-hit probabilities be always the same, no matter which side you play? And why so big difference between human/computer controlled player?

Does the game adjust difficulty to make it easier for the human player or does it report the hit-% wrong?

Voriax

[Wild Bill]

Variables are built into some many aspects of the game, Voriax, and for a reason. Results will change.

No battle ever fought twice under exactly the same circumstances is guaranteed the same outcome.

That is why it is so hard to recreate a "totally historical scenario" and expect the same results.

Maybe the gunner was scratching his butt, or smoking a cigarette, or chatting with another soldier.

Maybe he was fully attentive to his responsibilities. So many ambiguous actions in war can only be reflected by variables and they can sometimes drastically affect a battle.

Wild Bill

------------------
Wild Bill Wilder
Coordinator, Scenario Design
Matrix Games

[Mark_Ezra]

Correct me if I'm wrong....

There have been some discussion of low % of hit notification...2%, ect. SP shows two chance of hit %. The first is often very low. After you fire a second (actual) % of hit notification is shown. It usually is different than the first, although not always. Also the reverse can happen First % 75%!, take the shot...actual % 17%...It's better IMHO to pay alot of attention to the penetration tables than to the first (low) % notification.

[Voriax]

Bill, I get your point but that doesn't fully apply here. Variables are okay, at least those variables that've been discussed in other threads.

In my example however the variables seem odd. I mean, 10 times in a row exactly the same result. where is the variable? Why a 55% difference when the same team performs same action but under computer controlled side?

Just that it is not unusual in computer games to use a different set of rules for computer player than for human player, usually to cover up the lousy AI. Wonder if it's the same for SP series?

Mind you, I (still) think you have made the best SP game so far, it may just be the twisted mind of mine that ponder these things.

Voriax

ps. I once created a historical scenario for SP1, a Finnish STuG company counterattack during the 1944 russian offensive. I managed to get it playable for both sides, even winnable for both sides..historically it was almost a successful counterattack. Then my HD died and there went the scenario. But it managed to teach me what a pain scenario making can be.

[Paul Vebber]

Send me a save if you have it and I will look into it...