Matrix Games Forums

AI is a cruel cheater or I skip something in this life

Evidence:
1. Start new game:
"Typhoon" + "Axis Computer"
2. Wait when computer finished his turn, save
the game and exit.
3. Load this saved game with PLAYER="Both Human",
do empty soviet turn. Now we can have a look
into a computer cards: find 41st Panzer Korp
(near Rzhev), it constst of 7 units. For
example investigate the 19th Panzer Div.
Where it come from ?
4. Start new game:
"Typhoon" + "Soviet Computer" or "Both Human"
5. Again, find 41st Panzer Korp (near Velikiye
Luki) and examine its content. Well, well 8-o
here only 2 units instead of 7
and our 19th Panzer Div sit in the Center
Army Group in 11 hexes away from.

- AFAIK, it's impossible to plot Korp and
transfer into it new divs in the same turn
isn't it ? Derive valuable conclusions - computer
spit on the rules. :-/

This could be the 1st step in the machines taking over! <img src="eek.gif" border="0">

We better watch our appliances very closely over the next few days.................

[ November 22, 2001: Message edited by: Muzrub ]</p>

I can´t even remember one time the AI did´nt cheat.

Originally posted by pvv:
only 2 units instead of 7

---

The AI cheats horribly yet it is still an idiot, there is nothing else to say. It routinely overfills HQs and corps units. I once saw 8+ panzer divisions in the Finnish HQ, how they got there I don't know. <img src="smile.gif" border="0">

Originally posted by Ed Cogburn:

The AI cheats horribly yet it is still an idiot, there is nothing else to say. It routinely overfills HQs and corps units. I once saw 8+ panzer divisions in the Finnish HQ, how they got there I don't know. <img src="smile.gif" border="0">

---

Well, I see I dont discover the America here <img src="smile.gif" border="0">
Surely it possible to use some open source chess engine (GNUChess for example) to make AI more smart... but who have a time to do it.

well.. this will be the last one for now

will be a nightmare for arnaud to adapt any kind of program code to the game that only he can edit.

Originally posted by pvv:
Surely it possible to use some open source chess engine (GNUChess for example) to make AI more smart... but who have a time to do it.

---

Chess is a simplistic game. There is a small area, with a small number of pieces, all with a fixed capability and restricted movement. Chess programs generally succeed by using the brute force approach and they literaly test every move possible (with the really advanced chess programs of course it is more complicated).

A game like WiR would overwhelm any brute force approach program, so we can't look to chess as a possible solution.

Originally posted by SoleSurvivor:
will be a nightmare for arnaud to adapt any kind of program code to the game that only he can edit.

---

And according to Arnaud, there are just a few comments in the code, and no documentation on the AI code at all.

<img src="frown.gif" border="0"> <img src="frown.gif" border="0">

[ November 24, 2001: Message edited by: Ed Cogburn ]</p>

Originally posted by Ed Cogburn:

Chess is a simplistic game...
[cut]
A game like WiR would overwhelm any brute force approach program, so we can't look to chess as a possible solution.

---

Yeah, it's right. But Theory of Games is powerful science <img src="wink.gif" border="0"> Indeed (if we dont play against of Kasparov <img src="wink.gif" border="0"> ) we need three things:
<ul type="square">[*] function calculating the "price" of the position;[*] procedure for looking over all possible moves;[*] procedure for looking over positions in depth
and choosing the best continue;
[/list]The first two routines is a game-specific and are
used by the last routine that is almost "naked" theory and can be "cute-and-paste" from another step-by-step game. Yes, we never achive a grandmaster level with such approach. But defeating this AI will be not simple deal.
( and above all it will be honest fighting <img src="smile.gif" border="0"> )

anyway, source is not opened <img src="frown.gif" border="0">