 |
| |
|
|
Capt Dizle
|
|
I hope that Soren works on the pathfinding at some point.
Let me try to explain the problem.
Early in the game when the AI has only the tech to travel on coast squares you can easily block the AI's pathfinding by simply putting sea units in place so that the AI would have to travel in sea squares to get where it wants to go.
If you do this you can yo-yo the AI up and down your coast line. Block and watch the AI retreat, open a hole and watch the AI advance.
This is not good.
The same principal exists on land. Create strongpoints across a land mass and the AI will retreat in the face of it. Take care here to understand that the AI has been programed not to attack strongpoints.
Remove a single unit and the AI will advance directly to the hole in your picket line. Plug it, the AI retreats. Yo-yo.
The many ways to exploit this should be clear without further comment.
The above is not speculation, but is gathered by first hand experience. It is CERTAIN.
Had to throw that in for Cyclo....
I can't say right offhand how to fix this but I would love to see pathfinding and combat attack or do not attack routines randomized.
If culture flips can be dealt with in a manner as to yield a percentage per turn chance to occur, perhaps movement and combat inititives can be randomized to some extent. It would be an improvement.
|
|
| |
|
|
|
Capt Dizle
|
|
edited out.... 
I am wavering on whether or not you have to block the first sea tile (which you can do in certain circumstances).
Last edited by Capt Dizle on 19-06-2002 at 09:34
|
|
| |
|
|
|
DeepO
|
 |
supporting Candle'Bre
Jan 2002
time: 06:20
|
|
I observed this too, but as there is nothing Soren can do about it, I classified it an AI exploit, and stopped using it.
How I used it: in one of my games, I shared a long, banana shaped continent with English and Germans. As the English we're military stronger, I did not want to provoke a war. So when their side of the banana was fully settled, they wanted to cross my land to get to the area between me and the Germans.
I built a line of warriors from one side of the continent to the other, all in my territory, and so that no foreign troop could cross. So, I captured the English on their territory, as they didn't have galleys yet. But, at first I didn't have enough warriors in place, so wanted to keep those first two settler parties out as well. I formed a line of 4 warriors, and let them dance: move one tile up with the whole line, the settler would go down. Then I'd go down, the shortest way was up for the settler. I kept them dancing until I got the whole line in place -- could have been 50 turns or so, I was in no particular hurry until I was dancing around with 5 different settler parties. That became rather complex 
As a result, I growed to 3 times the size of England, keeping a lot of troops occupied. As I was building up my forces to finish England off, Germany(about the size of me) declared war to them. They just discovered map making. I figured it would be nice to keep everyone occupied in the middle of my territory, so I open up the line by removing one warrior, and let the opening dance around. From both sides troops came in, lots of troops. It was rather amusing to have knights, then cavalry, then even infantry dancing around my line of warriors. Meanwhile, I was gaining a big tech lead, and loads of tanks. So at a certain turn, I declared war to both England and Germany at once, mowed everything inside my territory down (used over 200 tanks, which all used up their movement points. Not bad for a standard map), and the next three or four turns were used in taking all of England and Germany. That was of course the easy part, there were only simple defenders left behind, as the AI had no chance of upgrading them with the maintenance for all the units they had running around.
After that, I'm still tempted to use the smaller version (4 warriors yo-yoing), but I try to resist. It is an exploit, no human player would fall for this, but there is no way an AI can counter it. The only possible solution is to attack the warriors to get through, or move everybody back home and ship them to the other side (which they would do if there weren't a gap in the line).
What was rather clever and unexpected, is that later on I tried to build up a full line (no gap this time) in a similar situation. One piece of advice: do not block the Russians when they are out for war. They saw the line, and without any warning declared war on me. And of course, in stead of battling the warriors, they went for my low defended cities... I never recuperated and had to abandon the game 
DeepO
|
|
| |
|
|
|
Coracle
|
|
I've seen the "yo-yo" nonsense occur with land units, also.
That has happened all too often with settlers marching into my territory; I block them, and they go south; I block them again and they had back north. We do that dance over and over.
I once did a TEDIOUS BORING dance with a stack of wandering Indian war elephants that kept moving into my territory from one of those stupid isolated towns that the dumb AI built deep in my territory.
|
|
| |
|
| |
|
|
|
DeepO
|
|
supporting Candle'Bre
Jan 2002
time: 06:20
|
|
Player1, I think that the path finding does prefer ships to get through instead of land. But, when an army has a land path of 3 moves to get to the other side of a small, 4 unit blockade, it will of course never walk 15 moves to a harbor, take a ship, and sail another 5 moves to get behind the line.
The fooling happens when a human player yo-yos, so that each turn the AI thinks the shortest landpath is 3 moves. But as you adjust the situation each turn, it stays 3 moves, instead of diminishing like in a normal path finding routine. That is the exploit, as said, in a static situation (blocking, and keeping put) the pathfinding algorithm will find a better solution, and if it can it will use ships to go around the blockade.
star mouse, if you had a static line instead of a moving one, that is just a tactic you could use, even in MP. After all, you just created a second line of defense, which is fine. It's when you start dancing, occupying the AI for a long time that it becomes an exploit.
Apocalypse: I think the path finding is quite good, given a static situation. The problem is that path finding is a relatively independent part of AI, which doesn't take into account the larger (political) situation. Therefore it doesn't notice that you are deliberately fooling it by changing the static situation each turn so that he has to yo-yo back and forth. You can do this in each TBS I know, and even in RTSes, although it is more difficult to perform there.
DeepO
|
|
| |
|
| |
|
|
|
Arrian
|
 |
Kneel before Grog!
Jul 2001
time: 00:20
|
|
quote:
Originally posted by nato
I do wish the AI would attack strongpoints a little more.
I have seen some very nice AI armies! I think it must have been very difficult to program the AI to build and concentrate all those units so well. But then instead of attacking my front line or even second line cities (which do have walls and some defenders) they parade past, trying to get to my inner, less defended cities. But they never make it! Often I make peace, and see the same impressive parade marching back home.
I think they might do better to just take a shot attacking. Just seems like it must have been harder to get the AI to build well and get the troops in a usable mass than to program it to attack with them. |
I concur. There have been several instances where an all-out assault on a key strongpoint of mine would have worked, and caused my great difficulty, but the AI refused to attack it, preferring to try and march around. Meanwhile, I was able to strengthen my defense, and start whittling away at their monster stack of troops.
The AI has got force concentration down pretty well. What it needs now is the killer instinct. I also think the AI needs a small tweak to redress the 1-move unit overkill it presently has. If the AI comes at me with 50 riflemen, I'm worried. If it came at me with 40 Cavalry and 10 riflemen, I'm terrified (unless, of course, I've got infantry  )
-Arrian
|
|
| |
|
|
|
ShredZ
|
|
Another thing about their 'all-mighty' stacks of troops is: when a unit takes 1 point of damage they head for home, no matter how deep they are in enemy territory. If they have a stack of 50 troops, and I have 15 bombarding units, I can usually send them all home before even one of them gets to attack me... oooh but that will all change with MP.
|
|
| |
|
|
|
Capt Dizle
|
|
Catt,
Given the fact that military conflict is not a shooting contest, but rather, a deadly dance, it is in fact impossible not to utilize your knowledge of how the AI will move.
Sure, I can avoid the more blatant exploits. But given the fact that the AI becomes so predictable in its movement and given the fact that the AI is hampered by the movement restrictions in enemy territory, the only way the human player can totally avoid taking advantage is to adopt random movement with his troops or intentionally move troops into bad positions.
I am speaking here primarilly of defensive actions. When the human is attacking this weakness is not so apparent.
What we need is more randomness in AI attacking movements, more willingness on the part of the AI to attack strongpoints. In essence Soren has programmed the AI to make the best percentage moves in most cases; however, once the human player can predict the moves, they are no longer the best ones for the AI to make. Therefore, randomness will often be more effective because it cannot be anticipated.
|
|
| |
|
|
|
Catt
|
 |
California - SF Bay Area
May 2002
time: 21:20
|
|
quote:
Originally posted by jimmytrick
Given the fact that military conflict is not a shooting contest, but rather, a deadly dance, it is in fact impossible not to utilize your knowledge of how the AI will move. [ . . . ] Sure, I can avoid the more blatant exploits . . . |
Utilizing knowledge of how you think the AI will move is one thing. But, I thought this thread was a complaint that the AI couldn't figure out you were playing it as a "yo-yo." I thought that my suggested solution to the yo-yo "bug" that you hoped Soren would work on was simple and elegant .
In any event, controlling your own behavior should not be impossible, even during the "deadly dance" of warfare. I know that the enemy AI at war will likely (not always, but very often) march many tiles deep into my territory to try and attack an undefended city. If I wanted to, I could deliberately leave an internal city undefended, and bombard the hell out of the AI stack for several turns before garrisoning troops in that city. I could (1) ask the programming team to try and devise a fix for this "bug," or (2) I could simply resolve to myself not to unfairly exploit the limited "I" of any AI.
This is really not that complex, and it's not a question of "knowing" or "suspecting" what the AIs next move will be. It is simply a matter of fighting the best tactical war that you can without taking actions whose sole purpose is to induce the AI to take action that, to HI, is clearly illogical.
Resist the urge to exploit . . . resist the urge to exploit . . .
Try it. It is really not that difficult.
|
|
| |
|
|
|
dexters
|
 |
Vancouver, Canada
Nov 2001
time: 05:20
|
|
Well Catt, some people can't help but cheat and then whine that they can cheat.
These are the same people who blame everyone but themselves for their own failings.
I've personally never used the undefended city, or worker bait exploit because I just don't use it. I prefer to divise my own strategy. Playing a game where I build a maginot line and watch my entire front light up with fireworks as streams of AI troopers march into my land is far for entertaining than any of the exploits. Last I checked, winning isn't why most people bought this game. Yes, they want to do well, and presumably win a few games. But the fun part is winning a game at the edge of your seat.
As a human, the AI can have all their faults fixed and we'll still have a massive advantage in brain power, not to mention we get to save and the AI don't.
This whole whining business is beyond me. But I suppose people feel important doing it. Whatever float their boat.
Last edited by dexters on 20-06-2002 at 09:53
|
|
| |
|
|
|
Capt Dizle
|
|
Catt, I just don't think you understand my point. When I am playing the game with almost certain knowledge of what the AI will do I am left with two choices...
1. Make a good move or..
2. Intentionally make poor moves.
Its that simple. It is not just a question of exploiting the AI.
I am simply hoping that Firaxis will consider putting a little randomness in there. I just don't want to be sure what the AI will do. I want to be forced to plan for more possibilities and have the possibility to be caught short when I fail to do so. It would make the game more fun.
Dexters, your comments are pretty interesting. In what way is your personal attack on me relevant to global warming?
I am a little disappointed in you. Still using the "save" exploit? Shame on you.
|
|
| |
|
|
|
Capt Dizle
|
|
quote:
Originally posted by dexters
Well Catt, some people can't help but cheat and then whine that they can cheat.
These are the same people who blame everyone but themselves for their own failings.
This whole whining business is beyond me. But I suppose people feel important doing it. Whatever float their boat. |
Dexters,
Would you care to explain at whom these remarks were aimed if not at me?
|
|
| |
|
|
|
dexters
|
|
Vancouver, Canada
Nov 2001
time: 05:20
|
|
I was responding to Catt's point and agreeing that people should exercise self control and choose not to use exploits --one example being the reloading save exploit. I then went ahead and noted that even if all these exploits are fixed, we still have an advantage so vast that in many ways, it doesn't matter. the AI can always be exploited, even subconciously. I mean, we all exploit AI every decision we make, from our decision to garrison units to how to screw the AI out of an extra dollar. Those are things you have to do. The Yo Yo effect is in many cases part of it, but from the subsequent discussion, it's been revealed some people openly exploit it, and that is where I draw the line and note that if they are going to whine about it being an exploit and asking firaxis to fit it, then should at least be capable of using self restrain and not exploiting it to begin with.
Yes, AI is predictable, yes, it can be exploited. But no, no one has to exploit it if they know they can. Perhaps the context of message was lost to you, but it is not an attack against you, if it was, i would be quoting your message and refuting your points.
|
|
| |
|
|
|
Capt Dizle
|
|
Personally, I respect people that have the guts to stand behind their insults.
|
|
| |
|
|
|
Capt Dizle
|
|
Dexie? You stand behind what? Your explaination or the non-insults?
You have me confused. At this point I am willing to concede your point if you'll concede mine.
I am a moron. And you are gutless. Agreed?
|
|
| |
All times are GMT. The time now is 05:20.
Apolyton Time is 00:20. |
top of page
|
| archivepost |
|
Forum Rules:
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
|
HTML code is ON
vB code is ON
Smilies are ON
[IMG] code is ON
|
|
|
|
|
|
19-06-2002 08:59
|