The Immortal Challenge 1: Apocalypto

My biggest objection to WFYABTA is the name. I played for a year, getting increasingly frustrated when AIs more advanced than me wouldn't trade. Finally, I was so far behind in one game that I didn't have the pre-reqs for the pre-reqs for the units I was seeing, yet I was "too advanced". That ticked me off enough to go sifting through old forums with the search feature, until I found a post that at least partially explained it.

If they'd just given it a name that reflected what it does ("We fear you are acquiring too much technology in trade"), I wouldn't have spent a year getting frustrated, and I could have started working on managing this important aspect of the game.

peace,
lilnev
 
I think the point is very well understood. The point is that strategy games are more enjoyable with a smarter opponent and a lower handicap, than using a dumber opponent and a higher handicap to achieve the same level of difficulty. It's pretty clear that Civ4 would be less fun if you made the computer opponents even dumber, and gave them even more free units and even larger discounts in order to make the game challenging. Civ4 is enjoyable, but it would be more enjoyable if you didn't have to give the opponents such huge bonuses (as in this present game) to create a challenge that keeps you interested in the game.

Correct. So why not change the handicaps? And change the other elements of bias that favour the AI right now? I'm not against better AI, and I say that for the hundreth time on this forum. I'm against better AI without corresponding modifications to the game rules for the sake of greater 'challenge'.

DaviddesJ said:
First of all, I want to repeat that it's not realistic to improve the AI that much, so I think you are exaggerating this concern.

Going by what some people have said about the Better AI project, I certainly did not cook this up. What they want is an AI that behaves like the most vicious human player in MP on steriods. To draw a historical parallel, did anyone exaggerate concerns about the Nazis and their intentions?

DaviddesJ said:
But, secondly, if that did happen, you aren't limiting anyone's options, because everyone can just stay at the "same level" of difficulty. As I said, players can go from old-Immortal to new-Monarch, from old-Monarch to new-Noble, etc. Everyone can have the same challenge, except for those who were already playing below Noble and may not be able to go any lower. But I would recommend "artificial stupidity" (intentionally worse AI decisions) for levels below Noble, which eliminates that problem, too.

Yes, you are limiting people's options in that they will be forced to rely on exploits or surefire methods to win above a certain difficulty level (eg. rushing the AIs). Instead of this happening on Immortal or Deity, it might be happening on Monarch on Emperor. Don't you think that is a negative development?
 
Correct. So why not change the handicaps?

By "change the handicaps" I assume you mean "change the names". I.e., the level at which the AI gets no discounts or advantages would be called "Monarch", instead of its current name "Noble".

As I said before, I don't care if you do change the names, and I don't care if you don't change the names. Since the names are just names, it makes no difference whether you change them, or not.

Going by what some people have said about the Better AI project, I certainly did not cook this up.

I have no opinion about the "Better AI project". I know very little about it. What I am responding to is your general statements about improving the AI.

Yes, you are limiting people's options in that they will be forced to rely on exploits or surefire methods to win above a certain difficulty level (eg. rushing the AIs). Instead of this happening on Immortal or Deity, it might be happening on Monarch on Emperor. Don't you think that is a negative development?

No, it's completely irrelevant. Whether the maximum level that you can defeat without "exploits" is called "Monarch", or called "Immortal", makes no difference. It's just a name. Why do you care so much about what the name is?

Consider the opposite case. Suppose that, in the next release of Civ4, Firaxis didn't change the AIs or handicaps at all, but changed all of the names of the difficulty levels, so that old-Noble would become "Settler", old-Prince would become "Noble", etc. Would you care? I don't care one way or the other. It's completely irrelevant, because they are just changing the names, and not anything that affects me when I actually play the game. You could edit the names back to whatever you want them to be called, if it really bothers you.
 
Uh... just so you know, your pictures aren't showing up(This is to Aelf), it says you exceeded your bandwith.
 
Anyway, I read VoiceofUnreason's thread on WFYABTA. It seems that WFYABTA applies between AIs as well but is affected by difficulty level handicap. Since the AIs are playing on Noble, I suppose they are applying Noble-level WFYABTA rules on each other while applying Immortal version on us. This still creates a difficulty-influenced bias against the human player. Maybe it's a necessary evil, but it's evil nonetheless ;)

I think I managed to confirm that the AI will stop trading with tech whore AIs (both by code inspection, and running a mod that reported the event).

Yes, the noble civs are being judged against the Noble threshold, which is 40% higher than yours; that means 2-8 techs, depending on who is doing the trading.
 
By "change the handicaps" I assume you mean "change the names". I.e., the level at which the AI gets no discounts or advantages would be called "Monarch", instead of its current name "Noble".

I think the handicap aelf is referring to is the production/research/promotion/etc bonus given to the AI.

Basically there are two ways to make the game more challenging: make AI smarter, or give AI advantages over human. aelf is arguing that since we are making AI smarter now with 2.08, we should take away some of the advantages, so that the game doesn't become one-dimensional.

I think we should take away the bonuses only if the "smartness" given to the AI is also a function of the difficulty setting like the bonuses. if that is the case, ie AI is smarter on Emperor than Monarch, then if we keep the current bonus setting, the AI is basically exponentially harder to beat at higher levels. However I don't think that's the case, the "smartness" upgrade in 2.08 should be uniform across all difficulty levels, and therefore the bonuses is still needed to differentiate the difficulty levels. maybe the bonus needs to be fine tuned so that Deity is "challenging" and not "frustrating," but there should be some bonus given to the AI at higher levels.
 
The Better AI team has released modified handicaps to try and balance the performance levels back to the original, by removing some of the more annoying crutches. For example, early Workers were removed or decreased in number to prevent the AI from stealing all the early wonders. The ridiculous unit upgrade discounts were flatlined at 50%, etc. Of course, now we have 1.61, 2.08, Better AI, and Better AI with new handicaps platforms ;)

Darrell
 
When I say I want a better AI--something I've been saying since Civ I--what I really mean is that I want the AI to stop doing stupid things. I think there's a thread around here that's filled to the brim with examples of what I'm talking about, and I think we all have our own pet peeves in this area. (My own include the AI leaving its siege weapons sitting idle as city defenders, stubbornly refusing to change its research or city builds to respond to an obvious threat to its very existence, and farming over its towns in the late game.)

I want to feel like I really achieved something when I win a victory (small or large) over the AI. I don't want to feel like I won because I took advantage of the computer's stupidity (or blind spots, whatever you want to call it).

I don't need or want the AI to behave like a vicious human from MP on crack. aelf's concern, and mine, is that currently, on the higher levels, you have no choice but to war with the AI almost constantly in order to win. We don't want to see that necessity spread to the lower difficulty levels where most players reside, as that would make the game one-dimensional. Currently, when I feel like a more peaceful build-oriented game, I drop a level. What if that was no longer an option? Or what if I had to drop to such a ridiculously low, easy level just to enjoy a different type of game?

When I show Civ to friends who are used to Age of Empires and the like, they're immediately impressed that it isn't just a war game. Warmonger that I've become, I'd rather not see the game go in that direction. I think it would lose a lot of its fan base and unique appeal.
 
Basically there are two ways to make the game more challenging: make AI smarter, or give AI advantages over human. aelf is arguing that since we are making AI smarter now with 2.08, we should take away some of the advantages, so that the game doesn't become one-dimensional.

And I'm arguing that that capability already exists. You can take away as much of those advantages as you want, just by going from old-Immortal to new-Emperor, or new-Monarch, or whichever of the new difficulty levels you prefer.
 
When I say I want a better AI--something I've been saying since Civ I--what I really mean is that I want the AI to stop doing stupid things.

aelf's concern, and mine, is that currently, on the higher levels, you have no choice but to war with the AI almost constantly in order to win.

If you make the AI do fewer stupid things, then, it's almost inevitable that human players will have to be more aggressive in order to win. Because war is one of the main things that humans still do better than AI players. So you can't have one without the other, to some extent.

You seem to, separately, be concerned about making the AI players more aggressive, as a way to increase the difficulty. I wouldn't favor that either. But that's quite different than many of the statements that aelf has made (although maybe he just wasn't expressing himself clearly).
 
If you make the AI do fewer stupid things, then, it's almost inevitable that human players will have to be more aggressive in order to win. Because war is one of the main things that humans still do better than AI players. So you can't have one without the other, to some extent.

You seem to, separately, be concerned about making the AI players more aggressive, as a way to increase the difficulty. I wouldn't favor that either. But that's quite different than many of the statements that aelf has made (although maybe he just wasn't expressing himself clearly).
Maybe I didn't express myself clearly enough.

No, I don't want the AI to be more aggressive--unless I am. If I choose to war against the AI, I'd prefer if it was a better war opponent that employed better tactics. (This is not to say there haven't been huge improvements here; I remember fortifying a Rifleman on a mountain in Civ II and watching as the AI lost unit after unit trying to displace him; at first it was funny, then it got tedious.)

Yes, war is something most humans are better at than the AI, but there are still other ways to overcome the AI without resorting exclusively to war. As I said earlier, I'd like to see the restrictions on early expansion throttled back a bit, which would allow the human player to have the option of peacefully obtaining the territory needed for a win--a change which has nothing to do with the AI.

Basically, humans are better at long-term strategic thinking than the AI, mostly because humans are better at holistic perception and thinking than the computer and probably always will be. This shows up most prominently in war, but it appears in other places too, like city specialization, tile improvements, tech and resource trading, diplomacy, and so on.

aelf himself has shown that there are other ways to win this game besides warfare; what he and I don't want to see is a scenario where the only way to win the game, at any level, is by warfare, and we're seeing a danger of that in some of the recent changes to the game and to the AI, especially when it's both smarter and has all these bonuses on the human player. Then it becomes hard to win with subtlety and necessitates picking up the ol' sledgehammer. Sometimes I'm in the mood for that (witness the current ALC game as Japan), but sometimes I'm not.

A smarter AI doesn't have to mean a more aggressive AI; that can be programmed into the various leaders' personalities. Frankly, the fact that the AI leaders do indeed seem to have individual personalities is one of my favourite features of Civ IV and is something I regard as an astounding feat of programming.
 
I hate to be the one to do this, since all your comments are interesting, but I'd rather move the discussion to another thread and leave this one for aelf's game only. We don't to swamp it so much that aelf won't be able to find the game-related suggestions, do we? :)
A good point, but I think aelf himself got the ball rolling and hijacked his own thread. :lol: We're all just impatient and eager to debate something while we wait for the master to sort out his PhotoBucket issues.
 
No, I don't want the AI to be more aggressive--unless I am.

That's what I was saying. That you are concerned (and/or aelf seems to be concerned) that the AIs might be made too consistently aggressive.

what he and I don't want to see is a scenario where the only way to win the game, at any level, is by warfare

I really think there's not the slightest danger of that. Maybe, at most, the AI could be improved enough that one must play more aggressively at Immortal, much as most people play at Deity now. That just means that new-Immortal is the replacement for old-Deity. There are still lots of lower levels. And, as I've said, if the AI gets too good (unlikely), you can always introduce artificial stupidity at those lower levels.
 
That's what I was saying. That you are concerned (and/or aelf seems to be concerned) that the AIs might be made too consistently aggressive.



I really think there's not the slightest danger of that. Maybe, at most, the AI could be improved enough that one must play more aggressively at Immortal, much as most people play at Deity now. That just means that new-Immortal is the replacement for old-Deity. There are still lots of lower levels. And, as I've said, if the AI gets too good (unlikely), you can always introduce artificial stupidity at those lower levels.


I don't think it's even a concern about making the AI's too "aggressive", really, but a concern that in making the AI play smart enough to provide a challenge to the Master players, it will force the human players at lower levels to play a strictly aggressive (warmongering) type of game just to keep up. Ultimately, it almost sounds like you and aelf/Sisiutil are arguing for the same goal from different directions. They'd like to see an AI that's smarter but doesn't get such a head start at high difficulty that you don't feel forced into a war situation just to bring things back to a more-or-less even keel, if I'm reading their posts right. You'd like the same thing, but it sounds like you would be open to (or maybe prefer, I can't quite tell) keeping some bonuses for the AI at high difficulty to increase the challenge level at the higher levels.

Personally I'd love to see an AI play as well as a human in all aspects of the game, including warfighting. But like you're saying, there would have to be some artificial limitations built in at lower levels because right now I doubt I could beat most of the "Big Guns" on here - so I'd be one of those people who would probably need a slight dumbing down of the AI for a while. :blush:

The funniest thing is I've been trying out Galactic Civilizations 2 (mostly last weekend but also an hour or so every night this week, the games can take forever in that game). They tout their AI as being highly intelligent and using "adaptive algorithms" instead of scripting to make the enemies more challenging. So I've spent most of the game translating in my head into Civ4 characteristics: Morale = Happy cap, Influence = Culture, etc. :D But ultimately I've seen nothing with the AI at its smartest that really impresses me. I would have been impressed if the AI had figured out that after the last war, I was far enough ahead that the only way to stop me would be to gang tackle me. So far, it hasn't. And just like the current Civ AI, the higher difficulty levels are only starting/research bonuses, not a smarter AI. I'm not sure what it is that makes a human so much smarter - the better flexibility, better predictive capability, or what - but one thing I do know is that there's a ways to go before the AI is thinking at or near a human level.
 
I don't think it's even a concern about making the AI's too "aggressive", really, but a concern that in making the AI play smart enough to provide a challenge to the Master players, it will force the human players at lower levels to play a strictly aggressive (warmongering) type of game just to keep up.

At lower levels, I still think it's a non-issue, because, if the AI ever does get that smart (which I doubt), some of the AI features can just be turned off for the lower difficulty levels.
 
The funniest thing is I've been trying out Galactic Civilizations 2 (mostly last weekend but also an hour or so every night this week, the games can take forever in that game). They tout their AI as being highly intelligent and using "adaptive algorithms" instead of scripting to make the enemies more challenging. So I've spent most of the game translating in my head into Civ4 characteristics: Morale = Happy cap, Influence = Culture, etc. :D But ultimately I've seen nothing with the AI at its smartest that really impresses me. I would have been impressed if the AI had figured out that after the last war, I was far enough ahead that the only way to stop me would be to gang tackle me. So far, it hasn't. And just like the current Civ AI, the higher difficulty levels are only starting/research bonuses, not a smarter AI. I'm not sure what it is that makes a human so much smarter - the better flexibility, better predictive capability, or what - but one thing I do know is that there's a ways to go before the AI is thinking at or near a human level.
I was actually going to bring up GalCiv2 as an example of what DaviddesJ is talking about with regard to dumbing down the AI on lower levels. This is actually what is done on the lower levels in GalCiv2. They disable some algorithm's in the AI depending on the difficulty setting. For instance, at some low-ish setting, the AI will fail to detect you amassing for a military strike. Turn up the difficulty a notch and you won't be able to get away with that without the AI noticing and countering. As it is, I believe the GalCiv2 AI isn't fully enabled until the top ~3 difficulties (though, I haven't played a new version of the game in about 6 months). After the AI is fully enabled, then they start giving it material bonuses to make it more competitive. The point being, for at least some of the learning curve, the AI is actually getting "better" alongside the player instead of just getting fewer penalties/more bonuses.
 
I was actually going to bring up GalCiv2 as an example of what DaviddesJ is talking about with regard to dumbing down the AI on lower levels. This is actually what is done on the lower levels in GalCiv2. They disable some algorithm's in the AI depending on the difficulty setting. For instance, at some low-ish setting, the AI will fail to detect you amassing for a military strike. Turn up the difficulty a notch and you won't be able to get away with that without the AI noticing and countering. As it is, I believe the GalCiv2 AI isn't fully enabled until the top ~3 difficulties (though, I haven't played a new version of the game in about 6 months). After the AI is fully enabled, then they start giving it material bonuses to make it more competitive. The point being, for at least some of the learning curve, the AI is actually getting "better" alongside the player instead of just getting fewer penalties/more bonuses.
I like the sound of this. I'd prefer this approach to increased bonuses only. This would give each level a very different feel, since some tactics you used on the previous levels would no longer work on the higher ones. Currently in Civ, with few exceptions, most of the tactics work on any level.
 
In GC2, the level at which the AIs use their full abilities is basically the equivalent of "Noble" in Civ4. There's no real reason to reduce their intelligence, because they aren't all that great (they are pretty much on par with the Civ4 AIs, in my opinion).

But the point remains. If someone does come up with a super-smart AI (unlikely), so good that even with minimal handicaps the AIs are very hard to beat, it would be easy to limit their intelligence below the highest level.
 
A good point, but I think aelf himself got the ball rolling and hijacked his own thread. :lol: We're all just impatient and eager to debate something while we wait for the master to sort out his PhotoBucket issues.

I did :lol: And you are absolutely right about why I did so. I'm 'exploring options' to solve the problem right now. I'll probably just upgrade to Pro.

I'm so tired of this debate, though. I don't mind doing so casually like on this thread, but I'm not going to participate in a full blown thread discussing this. In the end, there will invariably be a disagreement between people who don't like the idea of skewed difficulty levels and people who are perfectly fine with it (like what is happening now). To me, skewed difficulty levels encourages more cheese, which is not something I'd like to see in Civ4. Say what you want, but if even moderate difficulty levels requires you to play cheesily, people will play cheesily. And these people will claim to be experts at the game, while people like Sisiutil and me can't even run this kind of discussion threads anymore. What shall we call them? Warlord Challenge? All-Leaders' Noble Challenge? What authority would such discussions have in a community where cheese dominates?

Alright, I shall stop contributing positively to the OT discussion :p
 
Back
Top Bottom