Q: Custom AI in Warcraft2
Ren Wu
Hi,
Be a AI expert and the author of current best computer chinese chess
program and the author of a very strong chess program, I am also very
interested in making a strong WarCraft2 AI. I am already made changes
on the 'landattack' using WarDraft's buildin ai editor and the program
play much better game. (Beat the best WarCraft player in our company,
a gamer's company, who is also a very strong player on kali)
However, the things I can do by using the ai editor is very limited.
The fundamental weakness in the AI is still there, and no way to
change it. To make a super AI, I have to have better resource. So I
would like to know
1) is there any other tools available to make custom ai, apart from
the one in WarDraft
2) is it possible to make two computer ai to play each other? either
on same machine or on a local network?
3) if not, anyone have tried to setup 2 vs. 2 with one real human
player and a computer player in both side, and the humans just doing
nothing, will the ai play a resonable game?
4) is any infos available which tells how to interface with the
program?
5) is anyone thinking or have done a client program which can play
Warcraft game on Kali servers by its own?
6) Is there a standard protocol used between Kali server and the
client program? It is very easy to write a chess program which can
play chess game on a chess server, but I don't know what is the story
here.
I guess that's it. Thanks for your help.
Ren.
- remove one loop if you reply by email
LShaping
> Be a AI expert and the author of current best computer chinese chess
> program and the author of a very strong chess program, I am also very
> interested in making a strong WarCraft2 AI. I am already made changes
> on the 'landattack' using WarDraft's buildin ai editor and the program
> play much better game. (Beat the best WarCraft player in our company,
> a gamer's company, who is also a very strong player on kali)
Yeah, sure you did. I know a lot about Chess, a little about programming a
computer to play Chess, and enough about Warcraft. Unless you are an
infinitely intelligent being from outer space, you are not going to write a
good Warcraft AI. When it comes down to the details, there is little
similarity between Warcraft and Chess. Warcraft does not allow you to see
what your opponent is doing at first, unlike Chess. In Warcraft, you have
resources to gather. You are not limited to a eight by eight board. In
Chess, Knights can hop over other pieces, but that is nothing compared to
the fire of a tower, a ballista, an archer, or a mage. I will be glad to
discuss this further with you here in this newsgroup. And I am sure there
are many who would take you up on your boast about your AI beating a good
Kali player. I don't believe it.
@Zghoul
>I am already made changes
> on the 'landattack' using WarDraft's buildin ai editor and the program
> play much better game.
That is not very hard to do, just a few changes in the building order makes
drastic changes(like NOT going F F F Mill B), however, the computer only
gets that good, it can NEVER be compared to a good human player.
>(Beat the best WarCraft player in our company,
> a gamer's company, who is also a very strong player on kali)
In your dreams, only if your best player either sux, or the computer starts
out with a bunch of stuff, assuming of course that U have not tampered with
the building time/cost, of course noone will beat the computer if it
trains free grunts in 5 secs.
Could U not post this great AI in this group so we could try it out and see
if it rules so much?
--
@ZghouL13 - high resources are for newbies only!
LShaping
> Be a AI expert and the author of current best computer chinese chess
> program and the author of a very strong chess program, I am also very
> on the 'landattack' using WarDraft's buildin ai editor and the program
> a gamer's company, who is also a very strong player on kali)
Ren Wu probably posted and ran. If he wrote a very strong Chess program,
what is it called? Who published it? Where can we buy it? One thing he
should understand is the rules of challenge in a computer vs human game.
It was my e-mail that prompted IBM's sysop to include the rules at the site
of the recent match between world Chess champion Gary Kasparov and IBM's
Deep Blue computer. When humans play computers, the computer cannot have
control of the board. The computer must be prevented from cheating, like
@Zghoul said. Chess is no where near as complex or dynamic as Warcraft.
Sooner or later, someone will write a decent Warcraft AI. But not now and
not Ren Wu.
This is a commentary.
Copyright 1997 LSha...@DELETETHISibm.net
Ren Wu
On 18 Sep 97 21:15:20 GMT, "LShaping" <LSha...@DELETETHISibm.net>
wrote:
If you read rec.games.chess.computer, you know who i am. I understand
that you still at the the level of writing a chess program barely
generate legal moves, so there is not much I can tell you. We are not
in the same league!!! Understand?
A piece of advice, say something you know about. otherwise just shut
up!
Ren.
Chris Cothrun
In the interest of an intelligent conversation instead of flame wars
I'll raise the following points concerning AI versus Human play and
invite discussion. I've had some discussion on Kchat concerning this
topic and have my own opinions. When I mention the computer or an AI,
I'm not talking about the AI included with Warcraft, even doing
modifications via WarDraft. There just isn't enough smarts built into
the game as it is. Problems include pathing problems, position
evaluation, building placement, etc. I'm talking about a theoretical
AI with some intensive development behind it.
Reasons a Computer AI might be able to win at Warcraft:
Optimum Building Orders. The computer could analyze, develop, and
even change building orders on the fly that make maximum use of
resources. We all use some sort of building orders, whether it is by
the seat of our pants or following a list down to the 23rd peon. The
computer could execute this nearly perfectly, always having it's
resources spent and using them to the maximum effeciency. This
translates into hard grunt attacks and having ogres / knights and
magic very quickly.
"Multitasking" game play. This would be the computer's strongest
point. Continued production and expansion while attacking. Coordinated
multifront attacks. How is a human player going to be able to respond
to attacks multiple attacks at the same time? If he fends off the
first few is he going to be behind because he didn't keep peons in
production or didn't get the next couple of barracks started? Did he
get that second armor upgrade started as soon as the first was
finished? Little things like this are going to add up and grant some
advantage to the computer at various times. Imagine an opponent who,
while having equal numbers of grunts, had completed one of the armor
upgrades while you were still doing it. That would be a choice moment
to attack, even if it was a brief attack that only set you back a
little further. I think you get my point...
Human Player advantages:
Maps, Pathing and such: A human player, even a mediocre one such as
myself, is infinitely better at quickly analyzing a new map and
formulating rough stratagies for them. Is an AI optimised for POS
going to do very well on NWTR? Do we have to teach the AI how to play
each map or can we make it general enough to play good on all of them.
What happens when you hand it a new set of maps? This isn't chess,
where we can generate millions of combinations of 8x8 grids. Humans
can also get troops from one side of the map to the other without much
problem. I've seen quotes from programmers working on even the latest
games admitting that pathing, while greatly improved over War2's, is
still a problem.
Human Creativity: Whe can come up with some pretty great ideas. Is an
AI going to be able to react to everything we throw at it? that moves
us to my next item:
Flaws in AI: Witness Kasparov's game against Deep Blue where he opened
with a bizzare opening to throw the computer off of it's opening book
and I believe he went on to exploit this to a win in one of their
games. We are going to find quirks and flaws in any AI program, just
as the computer will find individual flaws in each of us. A good human
player will exploit these to his advantage.
These are about all I can think of and rationally explain right now.
What do you think? Are there more? How much of an advantage /
disadvantage would they be?
Chris
"crisco"
On 18 Sep 1997 17:25:52 GMT, "@Zghoul"
<jakob.o...@pc-programs.se> wrote:
>>I am already made changes
>> on the 'landattack' using WarDraft's buildin ai editor and the program
>> play much better game.
>
>That is not very hard to do, just a few changes in the building order makes
>drastic changes(like NOT going F F F Mill B), however, the computer only
>gets that good, it can NEVER be compared to a good human player.
>
>>(Beat the best WarCraft player in our company,
>> a gamer's company, who is also a very strong player on kali)
>
Adam Rizoviski
> control of the board. The computer must be prevented from cheating, like
> @Zghoul said. Chess is no where near as complex or dynamic as Warcraft.
> Sooner or later, someone will write a decent Warcraft AI. But not now and
> not Ren Wu.
>
>
> This is a commentary.
> Copyright 1997 LSha...@DELETETHISibm.net
(astonished voice)COMPLEX?! Com-plex?! You say Chess is no where near
cmplex as Warcraft?(astonshed voice over) Are you stupid? You may learn
how to play it but it takes years to master it. It is a highly
mathimatical game. It was created my a mathematition hundreds of years
ago. If you think Chess is simple, have you tried it? It is very logical
and involves complex stratagies. People have spent decades studying the
mathematics involved in the game. And personally, I don't think the people
at Blizzard are mathematitions.
Adam
--
Adam Rizewiski "I am Gul'dan. I will not be denied"
------------- Gul'dan the Warlock
"Mercy is weakness. And weakness is death"
Takhisis, The Dark Lady
@Zghoul
> If you read rec.games.chess.computer, you know who i am. I understand
> that you still at the the level of writing a chess program barely
> generate legal moves, so there is not much I can tell you. We are not
> in the same league!!! Understand?
>
> A piece of advice, say something you know about. otherwise just shut
> up!
Cute stuff, why don't you play him on war2 instead? If you don't have Kali
play over the zone, that is free. In the same league? Guess U are on top of
Cases ladder or National Warcraft League? I think he knows more then you
about war2, as stated earlier chess is not war2, and this is a war2 group,
if you are no good at multiplayer war2 you should not mouth off about him
not being in the same league as you.
--
@ZghouL13 - Chess is a boring piece of #¤&/ game, war2 rules eternal.
LShaping
Adam Rizoviski, I don't respond to insulting replies to my posts.
LShaping
Chris Cothrun, traditionally, programmers write for people, not for
computers. You talk about the bright future of programming Warcraft AI.
You suggest that Humans are forever limited to the present state of
Warcraft. I disagree. In tomorrows Warcraft, a Warcraft player, let's
call him Jack, will be able to upgrade weapons continuously and manage
resources optimally. Jack will be able to give multiple orders to a unit.
These orders will be carried out ASAP. Jack will be able to build a wall
easily. Jack will be able to keep peons in production. Jack says "Keep
peons in production!".
LShaping
john c fox, vaporware is good for bragging. Otherwise, it's useless.
john c fox
LShaping wrote:
>
> john c fox, vaporware is good for bragging. Otherwise, it's useless.
Not bragging.... Stated facts.
Oakie22
Thats exactly it, Chess is mathematical, thats why AI can be written for
it easier than warcraft2, Plus deep blue was a computer that did nothing
but run anything but chess algorythms. Warcraft 2's algorythms are much
more complex, and it isnt nearly as math orientated
Oakie22----http://members.aol.com/Oakie22/Warcraft2/mainframe.html
LShaping
> > call him Jack, will be able to upgrade weapons continuously and manage
> > resources optimally. Jack will be able to give multiple orders to a
> unit.
> > These orders will be carried out ASAP. Jack will be able to build a
wall
> > easily. Jack will be able to keep peons in production. Jack says
"Keep
> > peons in production!".
> What the hell are U talking about? Warcraft 3? war2 is finished U know,
> there will be NO more patches, or updates or versions.
@Zghoul, you need to review the thread.
LShaping
......
Ratty
I think maybe he means the AI of unit achieving it's objective and not
build/kill AI.... that would be stupid though since if you have UDTA,
u can actually increase a units AI....
Michael Garrison
to formulate chess in math for 100's of years
But the point is taken, WC2 is a lot more complex than chess
LShaping
> (astonished voice)COMPLEX?! Com-plex?! You say Chess is no where near
> cmplex as Warcraft?(astonshed voice over) Are you stupid? You may learn
> how to play it but it takes years to master it. It is a highly
> mathimatical game. It was created my a mathematition hundreds of years
> ago. If you think Chess is simple, have you tried it? It is very logical
> and involves complex stratagies. People have spent decades studying the
> mathematics involved in the game. And personally, I don't think the
people
> at Blizzard are mathematitions.
I never said Chess is simple. It does take many years to master Chess. I
feel the pain of Chess Masters who are no longer mentors. They have been
replaced by little boxes or by CD-ROMs. My most recent Chess experience
was at a Chess club here in a big city. For five years, I invented,
designed, and built many different digital Chess clocks. My best was one
for "bughouse" (some call it "bunkhouse"). Bughouse is the most popular
four player game of Chess. In bughouse, you play the opposite color of
your partner. So when you capture an opponent's piece, your partner can
use it. He can place the piece on the board, or keep it and make a regular
move. You can place a pawn up to the seventh rank. Some allow checkmate
by placing a piece on the board, some don't. Excuse me, got to go play
some Warcraft.
Spam Filter
In article <5vupq7$l...@freenet-news.carleton.ca>,
dw...@FreeNet.Carleton.CA (Adam Rizoviski) wrote:
:
: > control of the board. The computer must be prevented from cheating, like
: > @Zghoul said. Chess is no where near as complex or dynamic as Warcraft.
: > Sooner or later, someone will write a decent Warcraft AI. But not now and
: > not Ren Wu.
: >
: >
: > This is a commentary.
: > Copyright 1997 LSha...@DELETETHISibm.net
Actually, the built in AI for War2 is pretty formidable considering.
Few could beat it so handily if it were not severely limited in
post-keep smarts and, most of all, expansion AI.
: (astonished voice)COMPLEX?! Com-plex?! You say Chess is no where near
: cmplex as Warcraft?(astonshed voice over) Are you stupid? You may learn
: how to play it but it takes years to master it. It is a highly
: mathimatical game. It was created my a mathematition hundreds of years
: ago. If you think Chess is simple, have you tried it? It is very logical
: and involves complex stratagies. People have spent decades studying the
: mathematics involved in the game. And personally, I don't think the people
: at Blizzard are mathematitions.
:
: Adam
Warcraft is about click speed, not pattern recognition. The right
custom AI on a high-end workstation would beat a human player every
time.
--
David Poythress d...@qni.com
@Zghoul
> @Zghoul, you need to review the thread.
>
No
LShaping
> > Thats exactly it, Chess is mathematical, thats why AI can be written
for
> > it easier than warcraft2, Plus deep blue was a computer that did
nothing
> > but run anything but chess algorythms. Warcraft 2's algorythms are much
> > more complex, and it isnt nearly as math orientated
> > Oakie22----http://members.aol.com/Oakie22/Warcraft2/mainframe.html
> Sure, there are plenty of math involved in it. The damage system, the
> scoring system, and a buch of other stuff. I still think Chess is more
> complex. And Deep Blue did more then just algorythms. It supposedly
> thought more it self (exageration) and planned ahead. I could be wrong.
Not sure what "damage system" is supposed to mean. "This is the Chess
"scoring system".....a queen is worth 9, a rook is worth 5, a bishop is
worth 3.25, a knight is worth 3, a pawn is worth 1. By "A bunch of other
stuff", I guess you mean making legal moves and improving your position.
All Deep Blue did was think ahead. It though "What if?". If you read the
site of the match, you would have read that Deep Blue's strength was in
it's ability to calculate possibilities, the author called it "brute
strength". That is what Chess is. "If he does this, what will I do?"
Chess is a static game, Warcraft is a dynamic game.
G Wilson
Ren Wu wrote:
> >> Be a AI expert and the author of current best computer chinese chess
> >> program and the author of a very strong chess program, I am also very
> >> interested in making a strong WarCraft2 AI. I am already made changes
> >> on the 'landattack' using WarDraft's buildin ai editor and the program
> >> play much better game. (Beat the best WarCraft player in our company,
> >> a gamer's company, who is also a very strong player on kali)
Is there any chance that this land attack change could be made
public? I would *really* like to see the AI that could beat a "very
strong" player on KALI. Thanks!
G Wilson
--
IMPORTANT: When replying via email, please remove the "SPAMSHIELD." from
my email address.
@Zghoul
Ren Wu
Hi,
I post this reply three days ago, from earthlink's news server,
but it still not show on most other servers. So I try again, from
deja news this time. Sorry if you see repeat messages here.
Ren.
-----------------------------------------------------------------
On Fri, 19 Sep 1997 20:21:58 GMT, cot...@bigfoot.com (Chris Cothrun)
wrote:
First of all, thanks for your post. It make me feel a lot better.
This is really the usenet newsgroup for - to discuss ideas instead
flame each other.
>In the interest of an intelligent conversation instead of flame wars
>I'll raise the following points concerning AI versus Human play and
>invite discussion. I've had some discussion on Kchat concerning this
>topic and have my own opinions. When I mention the computer or an AI,
>I'm not talking about the AI included with Warcraft, even doing
>modifications via WarDraft. There just isn't enough smarts built into
>the game as it is. Problems include pathing problems, position
>evaluation, building placement, etc. I'm talking about a theoretical
>AI with some intensive development behind it.
>
Agreed. This is what I said in my orignal pos that the fundmental
weakness in the game's ai is unchangeable by only use the tools loke
wardraft. Instead, it got to be a totally new one.
path finding in the game ai is very bad, but I think this is easy to
fix. I think that in the current game, each unit find the path on its
own (on the fly) , it is fast but stupid. what we need here is a info
center (black board maybe) here and each unit can consult it when it
faced a path finding problem. if the center already have the info, the
unit just use it. if not, the unit then try find it way on it own. In
the mean time, the center will have a agent(a seperate thread with
special knowlodge of finding path) to search optimal path. Once it
find it, then notice all units. Even the finding path on the fly can
be improved significantly by let them do a few plies of search.
building placement, this has to be dealed at a few level. The basic
requiments here is not block the path between gold mine and the
townhall and so on. This can be handled by a few heuistics (rules)
at next level, it needs to know some patterns which gives best
defends, like build farms around the towers etcs. I think this can be
solved by using standard machine learning techniques, like genetic
algorithm. This is to say use it off line to find the pattern's value
and in the game ai, just use it. (you can think that is a super
building)
next level will be the knowlodge of overall structure of your whole
town. This is hard one and I don't have a very clear idea on it. I
have to think this a bit more.
To improve the ai, how to attack is also important. try
1. not to just send a few guys out. have a organized proper attack
with huge army.
2, try estimate wether or not the attack will succeed by using fliers
gather the infos.
3. try attack more than one palce at once, even though the ai know
that likily only one will succeed.
4. train ai to learn who should kill first.
I think all of these are not hard to do.
>Reasons a Computer AI might be able to win at Warcraft:
>
> Optimum Building Orders. The computer could analyze, develop, and
>even change building orders on the fly that make maximum use of
>resources. We all use some sort of building orders, whether it is by
>the seat of our pants or following a list down to the 23rd peon. The
>computer could execute this nearly perfectly, always having it's
>resources spent and using them to the maximum effeciency. This
>translates into hard grunt attacks and having ogres / knights and
>magic very quickly.
>
>"Multitasking" game play. This would be the computer's strongest
>point. Continued production and expansion while attacking. Coordinated
>multifront attacks. How is a human player going to be able to respond
>to attacks multiple attacks at the same time? If he fends off the
>first few is he going to be behind because he didn't keep peons in
>production or didn't get the next couple of barracks started? Did he
>get that second armor upgrade started as soon as the first was
>finished? Little things like this are going to add up and grant some
>advantage to the computer at various times. Imagine an opponent who,
>while having equal numbers of grunts, had completed one of the armor
>upgrades while you were still doing it. That would be a choice moment
>to attack, even if it was a brief attack that only set you back a
>little further. I think you get my point...
Agreed. See my idea above.
>
>Human Player advantages:
>
>Maps, Pathing and such: A human player, even a mediocre one such as
>myself, is infinitely better at quickly analyzing a new map and
>formulating rough stratagies for them. Is an AI optimised for POS
>going to do very well on NWTR? Do we have to teach the AI how to play
>each map or can we make it general enough to play good on all of them.
>What happens when you hand it a new set of maps? This isn't chess,
>where we can generate millions of combinations of 8x8 grids. Humans
>can also get troops from one side of the map to the other without much
>problem. I've seen quotes from programmers working on even the latest
>games admitting that pathing, while greatly improved over War2's, is
>still a problem.
I don't know what is their way to deal of this. I think they probably
fall into the trap that the path finding got be very fast for the real
time game and not using the best algorithm. Please see my suggestion
above. You need a special agent to find the best path and before he
find it, others just on their own. This way he will not lost any speed
and once the agent find the path, everyone else can just use it.
>Human Creativity: Whe can come up with some pretty great ideas. Is an
>AI going to be able to react to everything we throw at it? that moves
>us to my next item:
A few plies of search will give the computer the creativity, at least
the feeling of having creativite, if we don;t want go into the
pholisophy argument.
>Flaws in AI: Witness Kasparov's game against Deep Blue where he opened
>with a bizzare opening to throw the computer off of it's opening book
>and I believe he went on to exploit this to a win in one of their
>games.
sorry I split your paragraph into two. while I agree with you on next
setance, what you state here is wrong. GK won game one not because his
opening choise, instead he find a weaknees in DB's evaluation on
double connected pasepawns. He exploit it and win the game one. In
fact, he lost game 6 exactly because the bizzare opening he choosed!
> We are going to find quirks and flaws in any AI program, just
>as the computer will find individual flaws in each of us. A good human
>player will exploit these to his advantage.
Agreed.
>
>These are about all I can think of and rationally explain right now.
>What do you think? Are there more? How much of an advantage /
>disadvantage would they be?
I don't know if the ai can beat the best human players. I do think the
computer does have a considable advantage. However, I am confident
that a computer ai, which is knowledgeable, humanlike, fun to play
with, can be made and it is not very hard.
>
>
>On 18 Sep 1997 17:25:52 GMT, "@Zghoul"
><jakob.o...@pc-programs.se> wrote:
>
>>>I am already made changes
>>> on the 'landattack' using WarDraft's buildin ai editor and the program
>>> play much better game.
>>
>>That is not very hard to do, just a few changes in the building order makes
>>drastic changes(like NOT going F F F Mill B), however, the computer only
>>gets that good, it can NEVER be compared to a good human player.
>>
>>>(Beat the best WarCraft player in our company,
>>> a gamer's company, who is also a very strong player on kali)
>>
>>In your dreams, only if your best player either sux, or the computer starts
>>out with a bunch of stuff, assuming of course that U have not tampered with
>>the building time/cost, of course noone will beat the computer if it
>>trains free grunts in 5 secs.
>>
>>Could U not post this great AI in this group so we could try it out and see
>>if it rules so much?
>>
>>
>>--
>>@ZghouL13 - high resources are for newbies only!
>>
>>visit http://home3.swipnet.se/~w-38277/jakke
Whil I usually just igoring this kind of message, Chris's post put me
in a good mod to reply. Your message not show on my news server yet,
so I just reply here.
I am not keen to post any stuff here, apart from discuss some ideas. I
am not keen to prove anything, simply not worth it.
However, if you want to try, here is a few tips,
1. teach ai build second barrack ASAP.
2. teach ai build second town once you can defend your first one.
3. get enough footman before you upgrade
4. start attack ASAP and then launch attack in waves, each wave should
a litter bigger than last.
5. between waves, build more things or upgrade things.
6. have three townhalls after you have a keep and 4 if you have castle
7. build church and mega tower ahead others, (as soon as you can).
8. try to find a map the computer ai (not controled by ai editor) have
less trouble, I like friends.pud l alot.
Now you should get the idea that the build-in ai can still do
something. I know plenty of people they said that they learned
programming in c in a few days, but after few years they still write
Basci style or fortran style c program!
Ren.
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
LShaping
> > john c fox, vaporware is good for bragging. Otherwise, it's useless.
> Not bragging.... Stated facts.
.....The little dog laughed just to see such sport, and the dish ran away
with the spoon.
LShaping
Mr. Wu, you are saying the same things as did Mr. Cothrun on Friday,
September 19, 1997 15:21, the same things that I deflated with my post on
Saturday, September 20, 1997 13:54. You, John, and Chris are suggesting
that the computer will be able to do all sorts of wonderful things that
people wont be allowed to do. You are going to clear up the path between
the mine and the townhall. People have the same problems. Peons get
stuck. Troops get hung up on their way from here to there. Giving the
computer ridiculous advantages is OK for entertainment purposes. Good
Warcraft players do that already, playing seven computer opponents at one
time, and the computer loses pathetically , even though the computer AI
already has built in advantages. As I stated in my post on Saturday,
September 20, 1997 13:54, sooner or later, humans will be able to do these
things as well. You, John, and Chris sound like this. An analogy to
tennis. A tennis robot stands twenty feet tall, it can slowly move around
and serve the ball at 250 mph. The bot wins when it serves. It loses the
points that it doesn't serve. Here is another. You walk up to a video
game, put in your money, and the screen flashes "YOU LOSE!". That is what
your arguments sound like. If you want to argue about programming, take it
to another group. If you want to argue about Warcraft AI, I will.
@ZghouL13
> 1. teach ai build second barrack ASAP.
> 2. teach ai build second town once you can defend your first one.
> 3. get enough footman before you upgrade
> 4. start attack ASAP and then launch attack in waves, each wave should
> a litter bigger than last.
> 5. between waves, build more things or upgrade things.
What do U think I am? Fucking stupid? All of the above I have done, now U
must know that U cannot have to many "waves" in the beginning, 5 footmen
attacks(on my AI, the computer does not attack with less then 7 units
ever), means that the computer are so after on the upgrade, this does not
change the key thing thou, and please answer this even if U are not "in
the mood", the key thing is that if the human pull a NaTe formation or a
standard wallin, that computer run in with their guys and try to raze the
farms, when all the footmen are dead, one Grunt kills eventuall Cats(U
cannot time cats and grunts at exatct time, if U want a AI to work for
several levels/positions), and one Peons repair what was damaged, the
computer wastes tons of everything while the human upgrades early.
Yes a new AI makes it harder, my 7-man AI for medium pack hell of a good
punch, but if U just get up a good wallin(only needed against several
computers, unless the human suck), that attack is wasted.
> 6. have three townhalls after you have a keep and 4 if you have castle
Have not thought of this, but once the human have D&D all lines are DK-bait
anyway.
> 7. build church and mega tower ahead others, (as soon as you can).
U mean Altar and lust, that is what my AI went for, since lust is the best
spell in the game and the computer is a killer using it.
> 8. try to find a map the computer ai (not controled by ai editor) have
> less trouble, I like friends.pud l alot.
How about a pud that does not suck? Friends is so pathetic, 1000000000 gold
and one big open area, that is why newbies like it.
> Now you should get the idea that the build-in ai can still do
> something.
Against one computer your tips(above), that still makes the computer easy
bait, one NaTe and a couple of Grunts, and the Ogre rush, or why don't use
some a little more ultimate orders that U CANNOT learn the computer using
wardraft AI editor, how are U gonna learn the computer to defend against
Peon rush(Since U always know where the computer starts if U nkow the
level), ultimate FFB, rushes where U kill your Peons for food?
No_One
I am sick of seeing all these threads all over the place comparing Chess to
Warcraft, it is obvious that some of you have very little knowledge of what
is involved, or if you do, you do not bring it to bear here.
Chess is an intrinsically difficult mathematical problem to solve. It is
one of the mathematical problems classified as, NP-complete problems; the
Travelling Salesman Problem is a well known one. To explain very simply
what an N-P complete problem is, it means there is no KNOWN algorithm that
solves the problem and determines a winner any faster, than if you were to
systematically check all possibilities (ie. an exhaustive search) through
to checkmate for white or black.
It is the same thing PGP relies on to encrypt messages. ie. There is no
quick and easy way to break down a prime number generated with a 128bit key
without doing an exhaustive search for all possibilities. (That is one of
the reasons why the FBI and other government agencies hate it, because even
with a Cray-2 etc it is just too hard or too time consuming to justify, But
I diverge)
Now we have a chess board with, 8 x 8 squares. 32 x 32 is the smallest map
I am aware of in Warcraft II but I could be wrong.
In chess you have exactly the same number of pieces of exactly the same
type. In Warcraft II I concede that you probably have a similar amount of
pieces/men/monster things but they are not all exactly the same.
In chess all these pieces are more or less restricted to what they can do
and the only piece that is not restricted is of course the queen and we all
know how much we hate to lose her. In Warcraft II all of the characters can
move in any direction, hmmm, except when there is something in the way or
it is not allowed to go onto a certain square. (Geeze sounds like chess)
Now, taking in mind only the 3 points above.
Mathematically in theory chess is a simple game, because so far, we only
have a simple way to solve the problem involved.
I know I said before it was mathematically difficult, but I meant that it
is difficult to generate an algorithm that solves the chess problem. In the
end what we are left with, is effectively an almost infinitely large tree
diagram as the solution, which is obviously slow and cumbersome even for
the most powerful computers.
In practise though, even with its limited board size, number and movement
of pieces, chess quickly becomes a very complex game for a computer to deal
with because of all the possibilities that arise from each subsequent
possible move.
Warcraft II on the other hand has a larger board, with more pieces, which
have more possible actions they can perform and as the number of variables
increase the number of possibilities increase exponentially.
Now after saying all that, it is important to note what the objectives of
each game are.
Chess is about putting the opponents King in Checkmate. It has a definite
solution that leads to 1 of 3 possibilities (White or Black Checkmate or a
Stalemate) which so far is out of our reach technologically, does exist for
any given chess position.
Warcraft II's objective and a lot of other strategy games are to defeat
your opponent completely (sometimes). They are generally games of territory
and resource management ( He who haveth the Gold maketh the rules ) does
anybody disagree??. Yes I know in chess it is advantageous to have control
of the centre of the board but the centre of the board never changes. In
Warcraft II one position can start off being great but later on in the game
it can be a disadvantageous to have the same position.
Humans are very good at seeing the overall picture, while computers are
good at dealing with individual items one at a time albeit very quickly.
This is the crux of my argument and why I am sick of seeing Chess being
compared to Warcraft II and discussions about AI's applied to strategic
computer games.
On any particular map a computer could be programmed to play that map very
well, (I suspect that is how Blizzard programmed the campaigns) but stick
the same program on a different map and it would probably get slaughtered.
Sure you can give the computer some rules about where to build, and how to
place men and equipment, but that is what they used to do with chess
programs.
The chess programmers have given a lot (not all though) of that up because
the computer would follow the rule and if the rule no longer applied then
the computer was stuffed because it did not know what to do.
In Warcraft II the board actually changes as players build things and chop
down trees and mine out gold mines. Everything is fluid on the Warcraft II
board. This is why you cannot compare Chess AI's to a Warcraft II AI and is
also why it is very difficult for programmers to come up with a good
computer opponent for every given situation unlike a good human player who
will change his/her tactics to suit the conditions.
Oh well that is enough for me now hopefully we can get back to discussing
tactics again especially if we can get rid of that Mac V's PC thread. Dare
to dream. Hahaha J
-
If You can not find me, maybe it's because I am not here!
LShaping <LSha...@DELETETHISworldnet.att.net> wrote in article
<01bcc5f6$ef723fc0$997493cf@computer-name>...
Adam Rizoviski
Whatever. I am getting bored of debating this. You can debate by
yourselves. I like Warcraft, but I still think Chess is more complicated.
john c fox
Ok, no arguing...
Anyone know the flag that lets/tells the AI to build a cannon
tower???
LShaping
> Whatever. I am getting bored of debating this. You can debate by
> yourselves. I like Warcraft, but I still think Chess is more complicated.
I understand. Even though you are WAY off base about Chess being more
complicated. You aren't even on the right playing field. But, I know how
you feel. Try going to a Chess club and talking about how great Chess AI
is. And there, you are just telling the truth!
Ratty
>Chess is not mathematical in nature. Mathemiticians have been trying
>to formulate chess in math for 100's of years
They haven't done it yet?!
Ratty
I think the main problem with the AI of warcraft is that it is
unable to analyze the situation and attack accordingly.... They just
simply build the units, and set the units to "patrol" a certain target
in closet town (usually the townhall or barrak)
If they could somehow make the computer able to tell the
defense/ attack the position and sepearate them into pre-catergorized,
and follow the instruction of that catergory(my vocab sux, doesn't
it), then maybe it could be a bit more challenging then when they
continuously getting blown up by the same cannon turret one by one by
one by one.....
Michael Garrison
I also thought that the AI would attack the nearest "town". In a single
player pud that I was having a little trouble with I built a T.H. and
about 10 towers near it. They walk right past it (exept for the one that
actualy got hit by the towers) and continued for the main town.
Next try I did the same except that I place some popualtion with the
"town", Again main town got attact. I build "town" with T.H, and
population only This time they wiped it out.
I'v done a little experimenting and it seems that the AI does some
(limited) calculation at which town it would succede at attacking better
(limited because it quite otfen sends 6 or so to a 20 or so army)
It seems to wiegh peons a little higher than other population and
required building first (Lumber and Org mound over blacksmith )
(take lumber mill out and no trolls or balistas, take out blacksmith
after upgrading and no efect execpt no balistas, orgs still come out as
level 5)
if the AI did a little calculation after the attack (sent in 6, none
came back, killed only one) then it should know it has a problem with
it's first attack and at least dont repeat it (whether it can do better
the next time will depend on the AI's learning curve)
Ratty
>(astonished voice)COMPLEX?! Com-plex?! You say Chess is no where near
>cmplex as Warcraft?(astonshed voice over) Are you stupid? You may learn
>how to play it but it takes years to master it. It is a highly
>mathimatical game. It was created my a mathematition hundreds of years
>ago. If you think Chess is simple, have you tried it? It is very logical
>and involves complex stratagies. People have spent decades studying the
>mathematics involved in the game. And personally, I don't think the people
>at Blizzard are mathematitions.
GET OUT OF HERE YOU CHESS LOVING GEEK!
YOU probably haven't play war 2 either! We're not saying chess
is easy... just that it's MUCH easier to make a AI for.... because
every game... the AI can use the same strat over and over again with
different variables, but you can't do that in war2! there are probably
MILLIONS of puds out there and almost no 2 alike! YOu can't just make
a simple AI that can win a human oppenant in a hi resource map and win
him again in a map with no resource at all where it must win with
starting units!
LShaping
> YOU probably haven't play war 2 either! We're not saying chess
> is easy... just that it's MUCH easier to make a AI for.... because
> every game... the AI can use the same strat over and over again with
> different variables, but you can't do that in war2! there are probably
> MILLIONS of puds out there and almost no 2 alike! YOu can't just make
> a simple AI that can win a human oppenant in a hi resource map and win
> him again in a map with no resource at all where it must win with
> starting units!
Ratty the Oops, there is another basic difference between the two.
Warcraft type strategy games are in the early stages of development. These
new strategy games use the PC. The things built into the AI can be built
into the game itself, for us to use.
Adam Rizoviski
Ratty (ort...@tortoise.oise.on.ca) writes:
>>(astonished voice)COMPLEX?! Com-plex?! You say Chess is no where near
>>cmplex as Warcraft?(astonshed voice over) Are you stupid? You may learn
>>how to play it but it takes years to master it. It is a highly
>>mathimatical game. It was created my a mathematition hundreds of years
>>ago. If you think Chess is simple, have you tried it? It is very logical
>>and involves complex stratagies. People have spent decades studying the
>>mathematics involved in the game. And personally, I don't think the people
>>at Blizzard are mathematitions.
>
> GET OUT OF HERE YOU CHESS LOVING GEEK!
>
> YOU probably haven't play war 2 either! We're not saying chess
> is easy... just that it's MUCH easier to make a AI for.... because
> every game... the AI can use the same strat over and over again with
> different variables, but you can't do that in war2! there are probably
> MILLIONS of puds out there and almost no 2 alike! YOu can't just make
> a simple AI that can win a human oppenant in a hi resource map and win
> him again in a map with no resource at all where it must win with
> starting units!
I will have you know that I own all three Warcraft games. I figure myself
an novice player (definitly not a beginner) and I am pretty good in my
opinion. And if you don't have the IQ to play Chess, then don't and don't
critisize others for playing it. I my not be very good at it, but I do
enjoy playing it once and awhile.
Point is, don't critisize people unless you know them personally.
@ZghouL13
> I will have you know that I own all three Warcraft games. I figure myself
> an novice player (definitly not a beginner) and I am pretty good in my
> opinion.
The topic was AI, sure chess can be hard, but not for a computer, try
making a war2 AI that can play ALL levels and beat the top players, in
chess there is only one setup, say you started to play chess, with
different boards, thousands of them that is not similar, and give black and
white different playing pieces, that would make quite harder to make a AI
for it, would you not say?
> And if you don't have the IQ to play Chess, then don't and don't
> critisize others for playing it. I my not be very good at it, but I do
> enjoy playing it once and awhile.
>
> Point is, don't critisize people unless you know them personally.
Then why did you do it yourself(by saying he has a low IQ), in my opinion
it is not the smartest guys that play chess, it is the most bored guys.
George Weinberg
@ZghouL13 wrote:
opinion.
>
> The topic was AI, sure chess can be hard, but not for a computer, try
> making a war2 AI that can play ALL levels and beat the top players, in
> chess there is only one setup, say you started to play chess, with
> different boards, thousands of them that is not similar, and give black and
> white different playing pieces, that would make quite harder to make a AI
> for it, would you not say?
>
matter
when evaluating a middle game position. If you try writing a good chess
playing program, you will find it is not so easy. The reason chess
programs are
so much better than the W2 AI is simply that vastly more effort has been
put into
writing them, not that chess is a particularly easy game to write an AI
for.
George
LShaping
> Not particularly, no. What the starting position was like doesn't
> matter
> when evaluating a middle game position. If you try writing a good chess
> playing program, you will find it is not so easy. The reason chess
> programs are
> so much better than the W2 AI is simply that vastly more effort has been
> put into
> writing them, not that chess is a particularly easy game to write an AI
> for.
I am comming to the conclusion that the most important difference between a
Chess AI and a Warcraft AI is a fundamental one. Warcraft uses the
computer, Chess doesn't. The smarts that will go into a Warcraft AI will
be integrated into the Warcraft game for us to use. The game will evolve
as the AI evolves. The AI will always lag behind the game.
George Weinberg
LShaping wrote:
>
> I am comming to the conclusion that the most important difference between a
> Chess AI and a Warcraft AI is a fundamental one. Warcraft uses the
> computer, Chess doesn't. The smarts that will go into a Warcraft AI will
> be integrated into the Warcraft game for us to use. The game will evolve
> as the AI evolves. The AI will always lag behind the game.
I'm not sure what you're trying to say here. The part of the program
that askes "when should I upgrade to stronghod" or "should I try to
colonize this other island?" is pretty well seperated from the part
that askes, "how am I supposed to get to this gold mine? There
are trees in the way!"
Improving what I'd call the tactical intelligence won't necessarily
improve the strategic intelligence, nor vice versa.
George
LShaping
The game will get smarter. The input will improve too. Instead of giving
orders to individual units, you will have a subordinate managing the peons.
Knights will be ordering footmen around. Soon, you will be commanding an
entire kingdom, thousands of subjects. Still don't get it?
LShaping
LShaping
This double post happens when I go off line to write. So, I wont be going
off line to write anymore.
Gordon Wilson
In article <01bcce1e$8f36b840$b27493cf@computer-name>,
LSha...@DELETETHISworldnet.att.net says...
> The game will get smarter. The input will improve too. Instead of giving
> orders to individual units, you will have a subordinate managing the peons.
> Knights will be ordering footmen around. Soon, you will be commanding an
> entire kingdom, thousands of subjects. Still don't get it?
No. I don't get it. Seeing as Blizzard is done with WarCraft 2 and
has moved on to other projects (StarCraft, Diablo 2) who is going to be
programming all these wonderful new features into a game (albeit, a
fantastic game) that is already past its prime? Just curious.
Andrew Reisse
LShaping wrote:
> > > I am comming to the conclusion that the most important difference
> between a
> > > Chess AI and a Warcraft AI is a fundamental one. Warcraft uses the
> > > computer, Chess doesn't. The smarts that will go into a Warcraft AI
> will
> > > be integrated into the Warcraft game for us to use. The game will
> evolve
> > > as the AI evolves. The AI will always lag behind the game.
>
> > I'm not sure what you're trying to say here. The part of the program
> > that askes "when should I upgrade to stronghod" or "should I try to
> > colonize this other island?" is pretty well seperated from the part
> > that askes, "how am I supposed to get to this gold mine? There
> > are trees in the way!"
> >
> > Improving what I'd call the tactical intelligence won't necessarily
> > improve the strategic intelligence, nor vice versa.
>
> The game will get smarter. The input will improve too. Instead of giving
> orders to individual units, you will have a subordinate managing the peons.
> Knights will be ordering footmen around. Soon, you will be commanding an
> entire kingdom, thousands of subjects. Still don't get it?
Actually, improving the code that tries to find a path to somethings will
dramtically improve the computer's performance in many situations. For
instance, in gow, the green and purple positions may be able to actually do
something. Also, it would make the game more strategic in multiplayer. More
time would be spent on strategy than 'try to move a group of peasants to a new
mine'
Ratty
no I don't... what you're saying is simcity! what? knight ordering
footmen? I think that's going too far...
Nick Wise
>Ratty Ratty Ratty, you tell a knight to destroy a tower. The knight picks
>3 footmen and a cat. They destroy the tower. If you tell it to destroy
>the tower fast, the knight picks 3 footmen and two cats. If you tell the
>knight to destroy the tower carefully, it picks 3 footmen, a cat and a
>peon. The peon makes a defensive tower first, and the group destroys the
>tower from the safety of the defensive tower. What's the matter, can't
>handle it?
Yeah, and while your at it why don't you just tell your first peon to destroy
the enemy and he builds/attacks everything for you and wipes your ass while
he's at it. Come on, this seems to be taking a little of the interaction away.
LShaping
> no I don't... what you're saying is simcity! what? knight ordering
> footmen? I think that's going too far...
LShaping
> No. I don't get it. Seeing as Blizzard is done with WarCraft 2 and
> has moved on to other projects (StarCraft, Diablo 2) who is going to be
> programming all these wonderful new features into a game (albeit, a
> fantastic game) that is already past its prime? Just curious.
That's silly.
LShaping
> >Ratty Ratty Ratty, you tell a knight to destroy a tower. The knight
picks
> >3 footmen and a cat. They destroy the tower. If you tell it to destroy
> >the tower fast, the knight picks 3 footmen and two cats. If you tell
the
> >knight to destroy the tower carefully, it picks 3 footmen, a cat and a
> >peon. The peon makes a defensive tower first, and the group destroys
the
> >tower from the safety of the defensive tower. What's the matter, can't
> >handle it?
> Yeah, and while your at it why don't you just tell your first peon to
destroy
> the enemy and he builds/attacks everything for you and wipes your ass
while
> he's at it. Come on, this seems to be taking a little of the
interaction away.
Grow up.
LShaping
Nick Wise, I am describing a hierarchy. In a real kingdom, the king (which
is supposedly you, haha) doesn't even talk to a peon. The kingy give
orders to his subordinates, those subordinates give orders to their
subordinates, and so on. If a peon is in the way of a knight, the peon
moves. Catapults don't just sit out there in the open, they are protected
by footmen. Too easy? Not when you are managing thousands of subjects.
As these things are included, as the program becomes more complex, the AI
will be more difficult to write. Time to go play Warcraft 2.
Ratty
>Ratty Ratty Ratty, you tell a knight to destroy a tower. The knight picks
>3 footmen and a cat. They destroy the tower. If you tell it to destroy
>the tower fast, the knight picks 3 footmen and two cats. If you tell the
>knight to destroy the tower carefully, it picks 3 footmen, a cat and a
>peon. The peon makes a defensive tower first, and the group destroys the
>tower from the safety of the defensive tower. What's the matter, can't
>handle it?
SUre I can handle it... but some ppl just CAN"T, it would make
the game even more annoying trying to tell the knight NOT to take
footmen..... unless you make some kind of option thingy, but if you
have to do this to ALL the knights, it gets REALLY annoying..
Ratty
>Actually, improving the code that tries to find a path to somethings will
>dramtically improve the computer's performance in many situations. For
>instance, in gow, the green and purple positions may be able to actually do
>something. Also, it would make the game more strategic in multiplayer. More
>time would be spent on strategy than 'try to move a group of peasants to a new
>mine'
Eh? Ithought they said they made the peons stupid on purpose,
remember? the menu says that peons are supposedly dumber than all
units except ogres... I think they did that so you DO have to 'try to
LShaping
> > > > I am comming to the conclusion that the most important difference
> > between a
> > > > Chess AI and a Warcraft AI is a fundamental one. Warcraft uses the
> > > > computer, Chess doesn't. The smarts that will go into a Warcraft
AI
> > will
> > > > be integrated into the Warcraft game for us to use. The game will
> > evolve
> > > > as the AI evolves. The AI will always lag behind the game.
> > > I'm not sure what you're trying to say here. The part of the program
> > > that askes "when should I upgrade to stronghod" or "should I try to
> > > colonize this other island?" is pretty well seperated from the part
> > > that askes, "how am I supposed to get to this gold mine? There
> > > are trees in the way!"
> > >
> > > Improving what I'd call the tactical intelligence won't necessarily
> > > improve the strategic intelligence, nor vice versa.
> > The game will get smarter. The input will improve too. Instead of
giving
> > orders to individual units, you will have a subordinate managing the
peons.
> > Knights will be ordering footmen around. Soon, you will be commanding
an
> > entire kingdom, thousands of subjects. Still don't get it?
> Actually, improving the code that tries to find a path to somethings will
> dramtically improve the computer's performance in many situations. For
> instance, in gow, the green and purple positions may be able to actually
do
> something. Also, it would make the game more strategic in multiplayer.
More
> time would be spent on strategy than 'try to move a group of peasants to
a new
> mine'
Any improvement in the AI is a dramtic one. Improving the AI, as some
argue it, is improving the program too. The computer already has built in
advantages. It makes peons without pausing. We could do that, if we were
allowed. I am not at all sure what part "Warcraft uses the computer, Chess
doesn't." is difficult to understand. It's clear to me. The smarts that
can go into the AI can go into the game too. That is one distinct
difference between Chess AI and Warcraft AI.
Ratty
That just takes away all the elements of warcraft!!! It'll
turn into....... LORD OF THE REALMS!!! ARRGHHHHH!!!
George Weinberg
Ratty wrote:
>
> Eh? Ithought they said they made the peons stupid on purpose,
> remember?
Who said this? Where? I'm pretty sure nobody at Blizzard ever did.
> the menu says that peons are supposedly dumber than all
> units except ogres... I think they did that so you DO have to 'try to
> move a group of peasants to a new mine'
So how come mages aren't any better at finding their way around trees
than
peons are? They aren't, you know.
George
Ratty
On 6 Oct 1997 06:07:54 GMT, "LShaping"
<LSha...@DELETETHISworldnet.att.net> wrote:
>> Read the menu... ask anyone... they say so...
>
>
>
>
>Give it up, Ratty, how much brains does it take for group of peons to move
>thru a narrow passage? In your book, maybe a lot. In mine, not much.
ok.. you win.... I don't like to argue at all but that's what
I heard... so don't blame me if Blizzard or anyone else decides to
lie.... Because I didn't make this up! And pls don't reply to this
msg but cuz I accept that you made your point....
Dabeav9489
The computer uses a predefined algorithim for all land levels. While this
algorithm is changable through AI editors it is still a fixed algorithim. It
consists of a building order, rates for producing units, and orders to execute
in a linear fashion. While a little logic is actually programed into a fixed
portion of the Ai it is very limited. The comp builds farms, it may as you
suggest determine a favorable location to attack. Unfortunatly the AI does not
appear to have any learning curve at all. It attacks with predefined units at
predefined times. Not even a simple If then (a statement in programing to do
something if a condition is true. example, IF opponent army is smaller than my
army THEN attack) statement. basically the AI does NOT adapt to anything. It
does not seem to care what the player has done and will play the same way vs.
a taco bell strat as it does against air attack.
Jim Barlow
Dabeav9489 wrote:
The reason why you can't do this is because of complexity. The best strategies
are always the ones that exploit something more than intended, and rock your
enemy. However, exploiting is very dangerous. Taco bell strat (lots of sappers)
is very effective...
but not against an army of lusty ogres (not usually :). If the computer tries to
exploit something, it'll get screwed.
However, I do agree bliz didn't do great on the AI. It was good for a late 1995
game, but not great. It is linear, and could have used a "if ... then" command.
The problem with guys getting stuck just needs more time to compute the
destination. But at that time, it would probably have meant the minimum req. of
486/33 would have to be 486/66. Not much of a big deal now, but at the time they
would have lost a lot of market share.
But for one thing, the computer should have enough logic to repair the barracks.
Heck, it should repair everything it can. And the building order townhall farm
farm lumbermill farm barracks is just rotten. It was meant for any resources.
There should have been specific paths for specific resources
--
LordPeon
Jim_B...@bc.sympatico.ca
http://www.fortunecity.com/skyscraper/memphis/115
Jeff Tamer
I think what you guys mean my "stupid" is a mixture of cowardice and react
range. There are three levels of cowardice: Normal, Mage, and Peasant.
Normal cowardice is like any fighting unit, it will charge attacking troops
and attack if a troop walks near it. Mage cowardice means that it will
attack if a unit comes near, but will run when attacked. Peasant cowardice
means they could be standing right next to the enemy and wouldn't
automatically attack. They also run away when attacked.
React range is how far away an enemy is before the unit attacks it. For
example, maybe if you had two ogres, and one was killing some units 4 blocks
away then the other would come charging in and kill them too, but not at 5
blocks away (I just pulled those numbers out of the air). This makes ogres
look really stupid, like when there is a wall between them and archers. The
archers shoot at them and they run right up to the wall, trying to get at
the archers but of course they can't.
George Weinberg
Jeff Tamer wrote:
>
> >> Eh? Ithought they said they made the peons stupid on purpose,
> >> remember?
> >Who said this? Where? I'm pretty sure nobody at Blizzard ever did.
> >
> >> the menu says that peons are supposedly dumber than all
> >> units except ogres... I think they did that so you DO have to 'try to
> >> move a group of peasants to a new mine'
> >
> >So how come mages aren't any better at finding their way around trees
> >than
> >peons are? They aren't, you know.
>
> I think what you guys mean my "stupid" is a mixture of cowardice and react
> range. There are three levels of cowardice: Normal, Mage, and Peasant.
> Normal cowardice is like any fighting unit, it will charge attacking troops
> and attack if a troop walks near it. Mage cowardice means that it will
> attack if a unit comes near, but will run when attacked. Peasant cowardice
> means they could be standing right next to the enemy and wouldn't
> automatically attack. They also run away when attacked.
My impression is that they don't act this way. Peasants will fight back
when
attacked by other peasants, mages will fight fack when attacked by
death nights.
>
> React range is how far away an enemy is before the unit attacks it. For
> example, maybe if you had two ogres, and one was killing some units 4 blocks
> away then the other would come charging in and kill them too, but not at 5
> blocks away (I just pulled those numbers out of the air). This makes ogres
> look really stupid, like when there is a wall between them and archers. The
> archers shoot at them and they run right up to the wall, trying to get at
> the archers but of course they can't.
Knights will do this same as ogres. Archers aren't any smarter, it's
just that they can shoot
over the wall. They'll happily stand in catapult range.
George