Yet another AI CHEAT - Technology

[Willem]

Originally posted by Quokka


The sarcasm was noted and ignored. Chess is a more difficult and complex game than Civ.

And how do you justify that remark? Chess has a board comprised of 64 squares. I play on maps that are 256 x256 squares ( I'll let you figure out the math). Chess has only 9 distinct units in play at once (anyone tried a count of active units in the Modern Era?) Yet it takes IBM's most powerful supercomputer to even provide a challenge for a human Chessmaster. And if I recall correctly, the computer only won a single game. The rest it either lost or it turned out to be a stalemate. Feel free to provide the details and prove me wrong.

[Murtin]

Originally posted by Quokka
[The sarcasm was noted and ignored. Chess is a more difficult and complex game than Civ.

How? (Now I'm really curious of what you think of as complexity.)

Their research total might be sufficient but they shouldn't get the tech until the start of their next turn. The gripe is that the AI gets it or appears to get it on MY turn. The whole point of trading it to everyone on the same turn is to make sure you can sell it at all and avoid exactly what you describe from happening.

Whether they actually get the tech in the middle of your turn, or at the beginning of their own turn, doesn't make any difference at all as long as they don't do anything with it until their turn (such as giving it to a third party, for instance). Would you want them buy from you, even though their research total was sufficient? Now THAT would be a serious flaw in the AI since they'd be giving away something for nothing.

[Unregistered]

Humm. Two things going on.

1) Ai tech trading during the human turn. This is a known bug, recognized & acknowledged by Dan & Soren both. It will hopefully be addressed in the next patch...

2) Chess is more complex than civ. Sorry, not. The ai that IBM developed for big blue & the ai for civ (I assume anyway, I may be wrong on this...) in the main probably work the same way. They take a brute force approach and literally trial out every possible move and every possible response to that move for each movable piece (and combinations of pieces for a game like civ) out to a certain number of moves in the future. This is an incredibly complex process even for a game as limited as chess. Only 64 tiles, 32 pieces, & you can only move one piece at a time. Civ is a much larger board, the tiles themselves have different characteristics that alter the behavior of the pieces, I can have as many pieces as I want, & I can move all my pieces in the same turn. Not to mention that my pieces change abilities over time and you have to consider that there are up to 14 other players besides myself & the ai player whose turn it currently is that have to be accounted for. Also remember that Deep Blue was essentially a purpose built supercomputer with a team of the best programmers money can buy who were literally patching/'coaching' it throughout the match. Meanwhile, I'm running civ on my desktop...

Anyway,

[Quokka]

With Map Size don't let numbers confuse you. On a 256x256 map how many squares are active in your gameplay at one time? How many can you protect or threaten? Water can be pretty much totally ignored and that can be up to 70% of the surface. If you are not at war then that also disqualifies most of the AI controlled territory. So you really are dealing with a much smaller subset of the map than the whole 256x256. That greatly reduces the scope that you were trying to imply. Are all units active? Are any Automated? Large raw numbers aren't always an accurate determinator. The perfect example is the Army strength calculations in Civ3. An army of 100 Warriors appears larger and stronger then 30 Modern Armor but in reality.... Numbers are just that numbers. The possible applications of those numbers is what makes up complexity.
I just feel that all of the factors involved in Chess make it a more complex and dificult game than Civ. Strategies, subtleties and individual moves are more important in Chess than Civ. This is where I get the complexity. Civ has little room for subtleties or finesse, it is mainly a linear brute force force game. More of everything will win. More Armies, Research, Culture, Money and Cities will win it for you. Not so in Chess, how often has there been a conquest victory when the opponent still has more physical strength? In Chess you can manipulate you opponent so that the trap and effectively strangle themselves with their own pieces, not in Civ.
Because there is no way to directly correlate and compare the two games I guess my decision was based on a personal preference about each of the games. While there are more individual pieces available in Civ the ability to combine these into adhesive and succesful strategies is far more limited in my opinion than Chess. Factor in also the critical nature of each individual Chess move and this adds up to Chess being a more complex and challenging game in my book.

[Willem]

Originally posted by Quokka


Doesn't cover it at all. Sure the tech gets cheaper but why would they get it on MY turn? You only accumulate research on your own turn, same as production. You never produce or discover anything on a AI turn do you? Their research total might be sufficient but they shouldn't get the tech until the start of their next turn. The gripe is that the AI gets it or appears to get it on MY turn. The whole point of trading it to everyone on the same turn is to make sure you can sell it at all and avoid exactly what you describe from happening.

Check out this thread:

http://apolyton.net/forums/showthread.phps=&threadid=40594&highlight=Exploiting

The way I see it, if we can pull some of these stunts on the AI, it seems only fair that they have a few cards up their sleeve as well.

Regardless of whether Civilization is more complex than Chess, with our current technology we can't possibly hope to have an AI that can outwit us in the long run. It's stupid, we all know that it's stupid, and it always has been stupid since the first Civ came out. So if it has to cheat in order to provide us with a challenge, big deal. It's better than playing a game that's a foregone conclusion right from turn one.

[Murtin]

Originally posted by Unregistered
The ai that IBM developed for big blue & the ai for civ (I assume anyway, I may be wrong on this...) in the main probably work the same way. They take a brute force approach and literally trial out every possible move and every possible response to that move for each movable piece (and combinations of pieces for a game like civ) out to a certain number of moves in the future.

I'd say probably not. The search space for chess is huge, which is why you can only look a very limited number of moves ahead in trying every possible move and countermove. At some point you have to stop, and without further lookahead somehow evaluate the game state arrived at, in order to decide whether or not it's a good idea to go down that road.

With the size of Civ, I'm not sure it would be tractable to try out even one turn ahead, and one turn is really quite useless a lookahead anyway. On top of that, whereas in chess you have complete knowledge of the game state, in Civ you haven't. So you actually can't calculate the possible countermoves. And then you are left with the task of teaching the computer to make an "educated guess"... And even for a known game state, you'd have to design that evaluation function of a gazillion variables...

[Encomium]

STUPID AI

Tech cheats, and other cheats, are less bad than some of the dumb things the AI does.

1. Its civ is being seriously invaded, conquered even. So it starts building The Hanging Gardens! Or Oracle. Whatever. Instead of turning out military they end up destroyed.

2. It has a tiny civ alone on my continent. Said civ promptly allies itself with one across an ocean and decalres war on me. Of course, it is quickly exterminated.

3. Another civ is being conquered by me. I want to stop the War Weariness. . . but he refused three times in a row on different turns to see my envoy. So I finished him off so quickly I didn't even need to change governments. Dumb, fella.

4. A civ I had a Polite relationship with for a millennia suddenly declared war on me, even though outgunned, and thus sacrificed THREE luxury resources in trade. I took his one small city on my continent and easily started sinking his wooden ships with ironclads. He then sued for peace and it cost him all his gold, and the trade deal stayed dead.. Idiot.

5. The AI civ continued to send units (workers, settlers, military) across my borders to get to his few towns on the other side of my civ. I ordered him to leave the instant he entered my terrain, but he just kept going. War. Which he lost being outnumbered.

And on and on.

The AI civ strategy seems controlled by a mental defective.

THIS IS NOT FUN TO BEAT UP ON THE RETARDED.

[Willem]

Re: STUPID AI

Originally posted by Encomium

And on and on.

The AI civ strategy seems controlled by a mental defective.

THIS IS NOT FUN TO BEAT UP ON THE RETARDED.

I have yet to play a computer game that has the least bit of intelligence in it's AI, they always do stupid things. This is not a problem with the game design, it's a problem of the limitations of our technology. Like I said, it took IBM's biggest computer in order to provide even a short term challenge for a Chessmaster. Until we have computers as sophisticated as those seen on Star Trek, I guess you're just going to have to live with it.

[Murtin]

Re: STUPID AI

Originally posted by Encomium
The AI civ strategy seems controlled by a mental defective.

It's even worse than that. It's your computer.

[lawren8]

Re: Ocean Going Galleys - Not a myth

Originally posted by Deornwulf
I have seen ocean going galleys in more than one post-patch game I have played. I had the Lighthouse, so that couldn't be their excuse. I had Navigation, they did not. However, since I will not be installing Civ3 on my new machine, I will not be able to get the screenshots to prove it. (Besides, I have no clue how to get screenshots or screen captures.) Perhaps someone else can provide the evidence necessary to prove the existance of the ocean going galley.

I'm sick of seeing this complaint. Its not a cheat is a calculated risk.

I send my galleys out into the open ocean _all_ the time trying to
find the other continents. So what if a few galleys sink on the way -- the reward of finding the other continent first is more than worth it.

You should be happy that the AI is smart enough to try
a tactic that you haven't thought of before.

[muppet]

How can people even compare Chess AI to CivIII AI?

Chess AI is simpler to program than CivIII AI. Chess is more difficult to master for a Human than CivIII. Machines are not sentient beings -- some things are harder for them, and some things are harder for us.

It took how many high speed processors AND how many FIDE grand masters and programmers reprogramming the chess AI between games to beat Garry Kasparov?? The chess AI took how much time between moves?? The entire openning book chosen by the chess AI was CHOSEN by the grandmasters, not AI calculations. AI's even in chess can not 'think' strategically.

CivIII is a $50 game that runs on a PC. How much do we think it would cost to develop the chess AI that was used against Kasparov? How many of us could afford the processing power that beat Kasparov? How many of us would be willing to wait 20 minutes between critical moves per AI opponent? How many of us want to wait for firaxis or whoever to tweak the AI between games? How many of us can afford that level of service?

And even with all that said. The chess AI beat Kasparov because Kasparov was tired. The 'losing point' in the match game occured in a variation which Kasparov had played against Karpov during a world championship many years ago. That Kasparov made a mistake leading to his ultimate defeat could only be summed up as an oversight due to fatigue. What the media forgot to report was the shock and horror over the face of every professional chess player when Garry made the 'losing' move. They saw it instantly, so there is ample justification the the mistake was an oversight caused by fatigue.