Would people actually be interested in a Civ4 SGOTM?Methos said:Ok, dont even be joking about that! Ive been hoping they would start up a CivIV SGOTM ever since IVOTM came out.
Silly Prince
- Thread starter CB Droege
- Start date
akots
Poet
There might be problem with SGOTM as with any other SG before the final patch is released. People would have to keep multiple versions of Civ4 and the saves might be not compatible with these versions.
Same which happened with Conquests 1.15.
Same which happened with Conquests 1.15.
![[civ4]](/images/smilies/civ4.gif "Civ4 [civ4]")
. 
FriendlyFire
Codex WMDicanious
Prince is KILLING me.
Iam Dying here T_T
No seriously. Iam down to 10% science and I can only sustain this for a few more turns.
Iam Dying here T_T
No seriously. Iam down to 10% science and I can only sustain this for a few more turns.
Sounds more like you are killing yourself! 
Note the things you've done to get in this predicament, and don't do them in your next game!!
Note the things you've done to get in this predicament, and don't do them in your next game!!MyOtherName
Emperor
- Joined
- Dec 7, 2004
- Messages
- 1,526
You are vastly overestimating the abilities of this AI. And vastly underestimating just how much more simplistic chess is than CivIV.
CB Droege said:
In Chess, a position has on average 34 possible moves to consider. So to look two moves ahead, it's about 34^2 moves, and so on. So Chess computers can look a reasonable distance ahead, and beat the average player tactically (they can look even further ahead by carefully using choosing which moves to examine to the greatest depth). Chess computers get beaten by grandmasters because the grandmasters think strategically instead - it is harder for an AI to abstractly consider what's a good strategy than to concentrate on "what's a good move now".
In Civ4, a position has on average goodness knows how many hundred things to consider, so a simple chess-like search won't work. The AI can break off small chunks (point-to-point movement, order to send troops into battle, etc) and search for the most effective combination there, but for much of the game it simply has to rely on strategies and heuristics that have been coded into it by a human ("if I have more troops than him and he has some close weak-looking cities, maybe I should declare war"). Which means the AI has a fixed and fairly small repertoir to play from [small compared to a smart human, not compared to another AI]. And when the human learns how to counter a particular AI strategy, the AI is stuffed until the next patch. As a hypothetical but unproven example, as soon as you see an AI convert to Theocracy and Vassalage you know it's planning a war, so pop some high-strength troops into the cities of yours that it can see to encourage it to go after one of its other neighbours. If that counter works, then the AI's war-declaring heuristics have just been stuffed by the human to usually work against the AI.
CB Droege said:
I haven't played an RTS in 10 years, but I seem to recall that AIs in RTS games always have one unfair advantage over the human - they can give orders to two units on opposite ends of the map without having to move the "screen" over to the other units first. This speed advantage (not being slowed down by actually having to move a mouse, and being able to see the whole of the playing field at the same time) helps the AI no end. RTS games by their nature also tend to have fewer options at a given moment than turn-based games.
Not this again...
This comes back everytime, with the comparison with chess.
It's not the same.
Let's be clear.
If the best AI programmers in th world took 6 months to program a static AI for civ (meaning once its programmed, it doesnt change, even if it has some randomness in its decision process, the actual process doesnt change), it would still get TRASHED in a fair fight against a good human of the 'net community after maybe a few weeks of understanding the changes...
Anybody comparing civ to chess doesn't really understand how a chess AI works and how the same process applied to civ would fail miserably.
This comes back everytime, with the comparison with chess.
It's not the same.
Let's be clear.
If the best AI programmers in th world took 6 months to program a static AI for civ (meaning once its programmed, it doesnt change, even if it has some randomness in its decision process, the actual process doesnt change), it would still get TRASHED in a fair fight against a good human of the 'net community after maybe a few weeks of understanding the changes...
Anybody comparing civ to chess doesn't really understand how a chess AI works and how the same process applied to civ would fail miserably.
LulThyme said:
I find it strange that you moan about my contrasting the game with chess (actually in response to CB Droege bringing up the comparison), and then make exactly the same point I did about why it is different from chess and why a chess-like AI is not suitable.
He couldn't be talking about me... I do understand how a chess AI works, and I still stand by all of my points (even though you all seem to have missed the biggest point I was making).
Besides, You colectively talked me into playing the GOTM despite prince, and I was right, it wasn't fun. I gave it a real changce, I even enjoyed it for a few turns, but when I was slaughterd, I knew that it was only because the computer was given the advantage, not because it played a better game.
Whether or not it is technically possible to make the AI play a better game is entirely irrelevant to my points. My discussion in that realm was completely hypothetical.
Also, WarCraft 2 was the last game to really have the 'infinite mouse pointers' issue with the AI. Developers are aware of the issue, and have since built limits into the AI which address that problem. My comparison still stands:
RTS AIs are never made to 'cheat', like Civ AIs are.
I'm not trying to start an RTS is better than turn-based argument... I prefer turn-based games myself. I'm just tryint to say that I do not have the ability to have fun against a cheating AI, and I feel that this is not a flaw in myself, but a flaw in the game that only bothers a few fairness-oriented people.
This might be a better analogy than discussing chess and RTS and stuff.
If I enjoy running a good footrace, but I always win and want a greater challenge, what I want is to compete against a better runner.
I would not have fun if my 'greater challenge' was simply the same poor runner with a 15 meter head start
It may not be possible to find a better runner, but that's still what I want, and I would rather win everytime against the poor runners, than give any of them a head start.
The contest is meaningless if it is not fair.
If you want a greater challenge in Civ, it can be found in the settings, or in personal goals (my original exam[ple of a Duel sized world with 18 civs and 'diplomatic victory' the only possible win is very tough (even on noble, yes)) you could, for example, try to win by twice as many points as the second place civ, or you could try to get the space race victory a hundred years earlier. These are all ways of making the game tougher without making it unfair...
Besides, You colectively talked me into playing the GOTM despite prince, and I was right, it wasn't fun. I gave it a real changce, I even enjoyed it for a few turns, but when I was slaughterd, I knew that it was only because the computer was given the advantage, not because it played a better game.
Whether or not it is technically possible to make the AI play a better game is entirely irrelevant to my points. My discussion in that realm was completely hypothetical.
Also, WarCraft 2 was the last game to really have the 'infinite mouse pointers' issue with the AI. Developers are aware of the issue, and have since built limits into the AI which address that problem. My comparison still stands:
RTS AIs are never made to 'cheat', like Civ AIs are.
I'm not trying to start an RTS is better than turn-based argument... I prefer turn-based games myself. I'm just tryint to say that I do not have the ability to have fun against a cheating AI, and I feel that this is not a flaw in myself, but a flaw in the game that only bothers a few fairness-oriented people.
This might be a better analogy than discussing chess and RTS and stuff.
If I enjoy running a good footrace, but I always win and want a greater challenge, what I want is to compete against a better runner.
I would not have fun if my 'greater challenge' was simply the same poor runner with a 15 meter head start
It may not be possible to find a better runner, but that's still what I want, and I would rather win everytime against the poor runners, than give any of them a head start.
The contest is meaningless if it is not fair.
If you want a greater challenge in Civ, it can be found in the settings, or in personal goals (my original exam[ple of a Duel sized world with 18 civs and 'diplomatic victory' the only possible win is very tough (even on noble, yes)) you could, for example, try to win by twice as many points as the second place civ, or you could try to get the space race victory a hundred years earlier. These are all ways of making the game tougher without making it unfair...
CB Droege said:
AFAIK, RTS AIs get progressively discounted construction costs on harder difficultly levels. The original Red Alert was an example where the human player got cheaper units on easier difficulties.
And as for the AI not "cheating" on Noble, I did play one Noble game where two AI players discovered my continent. Two swift map trades later, and I could see the path that both took with their first caravels, straight to my island. In the original civ that exploit was allowed purely to make the game more interesting, if you started on an island by yourself.
CB Droege said:
I do not follow this argument against handicapping. You want a greater challenge, but would prefer to win every time rather than make things competitive. To return to the chess example, why not concede a piece head start? What makes things interesting is the relative progress. Yes, you learn more against better players, but that is because you learn more by losing than by winning. Just losing is less enjoyable. Your race would not be much greater challenge if you always won by 20m, but it would if you only ever won by 10m. And if I enjoy a good footrace, but lose every time, then it is not quite as good a footrace as one that I stand even a small chance of winning.
CB Droege said:
Indeed, by the end of your thread you are arguing for handicapping, but at the end of the game rather than the start. If you think it interesting to try to win by twice as many points as the second place civ, then why is it not interesting to win when the other civs have a 10% production advantage (etc)? Why is it more interesting to win by 10m than to start 10m behind?
As for your argument over the handicap slider, that is purely semantics. Obviously it is more difficult if you handicap yourself, and easier if you handicap the AI, whether by unit costs, or stopping the AI from using more advanced strategies or more blatant cheating.
CB Droege said:
Let's take your comparison with RTS.
Sure some of them dont make their AIs cheat.
But no good players play AIs because they kick their ass too much.
So the game becomes strictly multiplayer.
In Civ, they choose to avoid this by adding difficulty level.
You're talking about setting personnal goals.
Some people set the personnal goal of playing on Deity, which is a handicap, just as playing on a duel map with 18 opponents...
I don't know how far you want to take your footrace analogy, but beating a runner with 15 m headstart a beating the same with no headstart by 15 m (like your example with double the points) is pretty much the same in my mind,
akots
Poet
CB Droege said:
I'm sorry but it seems you are getting it wrong here.
Prince is a very low difficulty level strictly speaking and the game is not challenging absolutely on this level. There is no question human player would win imho if playing properly without mistakes. If you lose, you are obviously doing something wrong somewhere. Well, instead of figuring out what you are doing wrong and trying to make it right, you instead decided to complain about some essentially non-existent AI advatage. I'm really confused by this attitude. Like you admit voluntarily that you are not a good player and you do not want to improve. I'm sorry if I'm getting it wrong in my turn here.
MyOtherName
Emperor
- Joined
- Dec 7, 2004
- Messages
- 1,526
Judging from:CB Droege said:
"On the most challenging (and not actually possible to beat) level, the computer looks at every possible game outcome from every possible move, and chooses the move that gives him the best chances to win, negating anything you might try to do."
I beg to differ.
The computer only has the ability to look at "every possible game outcome" in a sparse endgame, and similar situations, where there are very few moves available. And even still, the computer could not achieve victory if it wasn't programmed with the (human) known algorithms for solving the basic end-games. E.G. I'm pretty sure that AIs are not yet capable of winning a king-knight-bishop vs king endgame with their searching abilities.
But the fact you said "and not actually possible to beat" is what most leads me to believe that you really don't know about this stuff.
But like most games, a good player can quickly get beyond the ability of the best RTS AIs (on a level playing field), and often beyond the ability of multiple AIs teamed up against him. In every computer wargame I've ever played, be it turn based or real time, the games are challenging at first, and then eventually become utterly trivial at anywhere near a level playing field.
One of my favorite wargames was actually based on this principle -- instead of making any semblance of an even fight, the whole point was that the AI had overwhelming forces, and the goal was to exploit (or create) a weakness in its position to achieve your objective.
Handicaps are just another dimension for making a game more challenging, and they have the advantage that handicaps can usually be scaled up indefinitely -- while you will eventually run out of new AI strategies, you generally will not run out of ways to give the AI other advantages.
Getting back to CIV -- think of it this way: if Noble was a fair challenge, you ought to be losing 86% of your games against 6 AIs.
The AI doesn't compete in the Games of the Month very often, so its ability to "cheat" is irrelevant. We have recently invited GOTM-AI to play some of the Civ3 Conquests games, taking the human role, and you can find his posts in the recent COTM spoilers. He failed miserably relative to the human competitors, losing all but one game. Only at Sid level did he manage to win a bronze medal, when he had the same advantages as the AI. This indicates the performance difference between the GOTM players and the AI.
For GOTMs the AI just provides a measuring stick against which all players pit their playing skills. It takes the place of the bunkers and hazards on a golf course, in a sense. Like bunkers, the AI don't cheat, they just sit there, getting in the way of progress. More so for novices than for experience players, which is the point, really.
We try to set the difficulty levels for games such that we can differentiate the best players from the good players, and the good players from each other. In a very easy game every competent player would get the same result, give or take a RNG roll or two. When the difficulty level is increased the AI gets more advantages, it's the same for all GOTM competitors, and different players cope with these challenges better or worse, creating an interesting performance table.
For GOTMs the AI just provides a measuring stick against which all players pit their playing skills. It takes the place of the bunkers and hazards on a golf course, in a sense. Like bunkers, the AI don't cheat, they just sit there, getting in the way of progress. More so for novices than for experience players, which is the point, really.
We try to set the difficulty levels for games such that we can differentiate the best players from the good players, and the good players from each other. In a very easy game every competent player would get the same result, give or take a RNG roll or two. When the difficulty level is increased the AI gets more advantages, it's the same for all GOTM competitors, and different players cope with these challenges better or worse, creating an interesting performance table.
solenoozerec
Stranger
AlanH said:
Why AI doesn't play 4OTM?
From reading your responces, I think the thing that really makes it a difference for me, is that I always play as if I'm playing against a person. I specifically try not to notice when the AI does things that I can predict, in order to make the game more enjoyable for myself... I don't think that it's fun to analyze an AI, and learn how to specifically work against the things it knows how to do. Would you be willing to play against a group of humans who all got their productions costs halved? poly not. It wouldn't be fun.
I'm not saying that I hate Prince because I always loose. It's that for some reason, on the occasions that I do loose, it upsets me more than it would if the computer didn't get overt advantages
The golf analogy even works... I would much rather strive to get better and better on a normal course, then try to keep the same level of performance on a course that with moving traps...
Again the point is not really to argue about chess AIs, as I said before, it was a hypothetical suggestion, but for the sake of argument: It is possible to program a computer with every possible chess game (there is a finite number of possible move lists). If you allow the computer as much time as it needs, it can simply choose the move each turn that keeps it on the most number of winning lists, and it will beat you. It doesn't matter how good you are.
I think despite the impossibility of winning against this hypothetical program, I wouldn't mind loosing, because it is at least following the same rules.
It could also be made more challenging (even impossible), by letting the computer move twice (for example) for each of your moves... That would not be fun, and you would not learn from it, because the AI is not playing the same game anymore.
However: what it really really comes down to is personal preference:
I prefer to play on Noble, and thought that there might be others who feel the same. If there were, I wanted to see if we could get a seperate GOTM.
It doesn't actually look like I have much support on that, so I've given up on the idea.
I'll just go play games that are solely against humans (which I prefer by far, anyway, despite the pains that this game puts us through for the experience)
I'm not saying that I hate Prince because I always loose. It's that for some reason, on the occasions that I do loose, it upsets me more than it would if the computer didn't get overt advantages
The golf analogy even works... I would much rather strive to get better and better on a normal course, then try to keep the same level of performance on a course that with moving traps...
Again the point is not really to argue about chess AIs, as I said before, it was a hypothetical suggestion, but for the sake of argument: It is possible to program a computer with every possible chess game (there is a finite number of possible move lists). If you allow the computer as much time as it needs, it can simply choose the move each turn that keeps it on the most number of winning lists, and it will beat you. It doesn't matter how good you are.
I think despite the impossibility of winning against this hypothetical program, I wouldn't mind loosing, because it is at least following the same rules.
It could also be made more challenging (even impossible), by letting the computer move twice (for example) for each of your moves... That would not be fun, and you would not learn from it, because the AI is not playing the same game anymore.
However: what it really really comes down to is personal preference:
I prefer to play on Noble, and thought that there might be others who feel the same. If there were, I wanted to see if we could get a seperate GOTM.
It doesn't actually look like I have much support on that, so I've given up on the idea.
I'll just go play games that are solely against humans (which I prefer by far, anyway, despite the pains that this game puts us through for the experience)
solenoozerec
Stranger
CB Droege said:
Why not? It depends on their skills. If they play considerably worse than me, I'd prefer to equalize the difference. It is not fun to play when you know the outcome.
well... in gotm you usualy know the outcome, but you do not know how you will do compared to other players. Thus, it is still fun.
Similar threads
- Sticky
