What needs improvement in Civ IV?

[nbarclay]

Originally posted by Randolph
In terms of simply being attacked all the time, I’ve got no problem with that either. Part of being a good builder is making a strong defensive network. I wouldn’t want to push it so far that the AI will attack you no matter what, one of the huge advances in Civ4 is [relatively] meaningful diplomacy, particularly regarding religion. This certainly isn’t the case now: even in my games with ~1.5(normal for map size) civs and aggressive AI on, I’m pretty safe from my “brothers and sisters of the faith” well into the game.

And what happens if you founded one religion and your neighbor founded another? Personally, I already find myself getting attacked when I'm not in a mood to fight often enough that I don't especially care for the idea of making the standard AI behavior more aggressive.

[Randolph]

Originally posted by nbarclay


And what happens if you founded one religion and your neighbor founded another? Personally, I already find myself getting attacked when I'm not in a mood to fight often enough that I don't especially care for the idea of making the standard AI behavior more aggressive.

I don’t think were disagreeing about the facts: if the AI was more aggressive the game would be more difficult without giving the AI any mathematical bonus. If you’re next to a civ with a different state religion there is going to be a lot of war, and life is going to be generally more difficult for a builder. But you’re not without strategic options (while maintaining a basically builder style): (1) you could spread your neighbor’s religion, and convert; (2) you can try to cripple them through war and diplomacy; (3) you can set up a strong defense. When this happens I don’t feel like I’m being railroaded into a certain game dynamic like I do with the demand system; there is a reasonable cost-benefit trade off.

Edit: I'm in full agreement with your post RE the demand system.

[nbarclay]

Originally posted by Randolph

I don’t think were disagreeing about the facts: if the AI was more aggressive the game would be more difficult without giving the AI any mathematical bonus.

My suspicion is actually that greater AI aggression would have mixed results. In Civ 3, successful conquest was a vital ingredient in creating "killer AIs" because all else being anywhere near equal, a large civ always had a stronger economy than a smaller one. But in Civ IV, even successful conquest can at least temporarily wreck a civ's economy. I remember one game where Napoleon (if I remember who it was correctly) wiped out the other civ on his continent (Spain I think) and ended up with almost no GNP. French technology never did become particularly competitive. (I might mention that this was in one of my huge-map 18-civ games).

So what we have in Civ IV is a situation where unsuccessful warfare against each other wastes AIs' resources and where even successful warfare may or may not make a victorious AI stronger. That leaves me skeptical as to whether increasing AI aggression would result in a reliable increase in difficulty - at least unless and until the AIs get more reliable at leveraging successful conquest into strong technology.

[Randolph]

Originally posted by nbarclay


My suspicion is actually that greater AI aggression would have mixed results. In Civ 3, successful conquest was a vital ingredient in creating "killer AIs" because all else being anywhere near equal, a large civ always had a stronger economy than a smaller one. But in Civ IV, even successful conquest can at least temporarily wreck a civ's economy. I remember one game where Napoleon (if I remember who it was correctly) wiped out the other civ on his continent (Spain I think) and ended up with almost no GNP. French technology never did become particularly competitive. (I might mention that this was in one of my huge-map 18-civ games).

So what we have in Civ IV is a situation where unsuccessful warfare against each other wastes AIs' resources and where even successful warfare may or may not make a victorious AI stronger. That leaves me skeptical as to whether increasing AI aggression would result in a reliable increase in difficulty - at least unless and until the AIs get more reliable at leveraging successful conquest into strong technology.

I agree. That’s why in my first post I suggested decreasing city/civic upkeep costs for the AI (on higher levels). The idea is that this creates a latter-game bonus to the AI opposed to the, extremely annoying, early bonuses (e.g. the free worker). This would help keep the AI competitive in the latter game, while not making every game be one of “catch up.” In this same vein I also proposed making it easier for an AI to get another AI into a war (this appears to be somewhat implemented in Warlords), in order to reduce the number of no-winner wars that simply serve to drain the AIs of resources.

I think I’ll try to experiment with this a little in my next game by selecting aggressive AI, and choosing a number of organized civs as opponents.

[fizz]

One thing that i always found unrealistic in all civ games and that could help reduce problem n°1 is how big the technological advantage usually become in later stages of the game (this is at least true for me, a builder by nature). The internet help reduce this problem, at least respect to the old civ3, but i think that's introduced a bit too late in the game for having any effect... a civ that can buy it in the first place, does not have many troubles keeping up with techonlogy. Maybe, introducing a weaker version of it (for example requiring 3-4 civs, or introducing a time-lag), but introducing it a lot earlier, could help making the game interesting in late stages...

[V3nom]

perhaps I'm looking at things too simply here but everyone thinks that harder AIs get too many early advantages and lag behind later on (and I agree) so wouldnt the simplest solution be to give the AI less advantages at the start and have the advantages scale as time goes by?

Even possibly have it scale so the more powerful the player is the more advantages the AI gets (not to the point where it completely counter-acts the advantage in getting more powerful, but enough that it increases the challenge slightly)

i.e. the further ahead a player gets in technology the bigger the percentage boost to research given to the AI. The larger your military in comparison to the average military force, the cheaper units are for the AI. These advantages would of course scale according to difficulty, at chieftain they would be almost un-noticable while at deity it'd mean that it'd be a constant challenge for much of the game to solidify you're advantage (although of course if your dominance was great enough it would have to still be possible).

[DrSpike]

I'm not sure if I like that idea. It seems to undermine the incentive to play well.

[Kloreep]

Originally posted by V3nom
perhaps I'm looking at things too simply here but everyone thinks that harder AIs get too many early advantages and lag behind later on (and I agree) so wouldnt the simplest solution be to give the AI less advantages at the start and have the advantages scale as time goes by?

Interesting proposal to me. Could the AI advantages be scripted to change this way? I'd guess even this could be done, I haven't looked into the things that can be changed with Python but it sounds pretty powerful...

This does have some potential for abuse, though. Particularly, I think this would favor early rushes, since you would effectively be rushing a civ at say, Monarch level if you attacked early on in the game, but might be facing a Deity level AI if you let time pass before going after the same civ. So the amount of bonuses not given to the AIs at start would have to be carefully balanced; too much difference between the start and end of a game, and aggressive early-game strategies would probably be encouraged.

Originally posted by V3nom
Even possibly have it scale so the more powerful the player is the more advantages the AI gets (not to the point where it completely counter-acts the advantage in getting more powerful, but enough that it increases the challenge slightly)

This would be very tough to balance so that it wasn't overwhelming. Personally, I hate "rubber band physics" in games. If the AI has to cheat to compete, fine, but I want it to do so irregardless of how I'm doing. Like DrSpike says, it takes away much of the reward for success, and can even make it feel like it is better to not do so well and simply try to slip under the radar.

[nbarclay]

Originally posted by Kloreep

This does have some potential for abuse, though. Particularly, I think this would favor early rushes, since you would effectively be rushing a civ at say, Monarch level if you attacked early on in the game, but might be facing a Deity level AI if you let time pass before going after the same civ.

The trick is to set things up so that builders will have rough technological parity with the leading AIs throughout the game if they play on a difficulty level that seeks an even match. Then fighting early won't be necessary to keep Monarch level AIs from turning into Deity-level opponents, but it won't generally be possible to build up a tech lead that enables previously peaceful players to stomp all over the leading AIs with minimal effort later on.

This would be very tough to balance so that it wasn't overwhelming. Personally, I hate "rubber band physics" in games. If the AI has to cheat to compete, fine, but I want it to do so irregardless of how I'm doing. Like DrSpike says, it takes away much of the reward for success, and can even make it feel like it is better to not do so well and simply try to slip under the radar.

For the most part, I strongly agree that giving AIs extra bonuses if the player is doing well takes away the rewards that rightfully belong to players who play well. But I do think it would be useful if the research system gave additional discounts to backward civs. That would make it harder to have ridiculously lopsided situations like modern armor killing riflemen.

The ideal would probably be to have discounts come into play not just when other civs research a tech but also when other civs research successor techs - either techs that the tech in question is a prerequisite for or techs that are a lot later in the tech tree. Then the more primitive a tech becomes compared with the state of the art and with the general state of the world's technology, the bigger the discount would be. But setting up such a mechanism would be a whole lot more complex than editing XML files, assuming there's a way the modding community can do it at all.

[V3nom]

If set up incorrectly yes it would take away the reward for success. However I think that if it was EXTREMELY carefully balanced (with much playtesting) it could work, but only if the players success is only one factor, not the only factor, to determine the AIs advantage.

Diety level AI should still be able to out-produce you early game (although perhaps not to the same extent if they're going to get more advantages later on). Cheiftain level AI should still have little chance of competing with you, perhaps to the point where even if their advantage was maxed their units and techs would only cost the same as yours (and of course since you'd have more cities you'd still be progressing at a faster rate than them).

I also dont believe this would encourage early rushing. For two reasons:

One, The AI would still get some advantage/disadvantage initially depending on the difficulty you choose, independant of what I am proposing

Two, Since it scales a huge military when compared to the AIs early on would give the AI a discount on its units the same as a huge military later on. You'd have to give the AI bonuses based on a ratio of your power (in military or technology) compared to the AIs, to do it with predefined threshold which are absolute values would be doomed to failure.

Also in regards to the reducing of rewards for success: If badly implemented then yes this is absolutely what would happen. Infact this is probably what would happen if you released a beta modification that did this until HUGE amounts of playtesting can be done to balance it.However I believe, if correctly implemented this wouldnt be an issue.

The key would be to have some point in any game in which globally you are equal to the average AIs level, you can produce and research exactly as fast as them. Before this point you will be at a disadvantage (despite the AI getting little bonus from the power comparison) and after this point you will still outpace them (despite their relatively high bonus' from the comparison).

In doing this you would make it so that early on, you still have as much incentive to grow as you ever do (because if you dont reach that threshold, you're not going to win are ya?) and after you achieve it you still have incentive to gain more power because you will be pulling ahead of the AI. It is no different from the current system in that respect, the difference would be in how long the game remained interesting after you surpass parity with them (which for most games, is very short).

As for wether its possible to be scripted in I'm not sure, having not done much civ4 mod work. However my guess is that changes of this magnitude would require editing of the source code.

edit: In regard to the 2nd problem mentioned in the OP, why not just make it so the extra happiness granted to each city is worked out by comparing the number of military units you have within your borders to your total number of cities?

[Nacht]

Originally posted by Randolph
I agree. That’s why in my first post I suggested decreasing city/civic upkeep costs for the AI (on higher levels). The idea is that this creates a latter-game bonus to the AI opposed to the, extremely annoying, early bonuses (e.g. the free worker).

This can be done by modifying the iAIPerEraModifier in CIV4HandicapInfo.xml.

This variable gives the AI additional bonuses per era. It's set at 0 on all levels up to noble and -1 on prince to -5 for deity.
It effects all building costs, city/civic/unit upkeep costs and war weariness. It doesn't directly influence research costs.

So for example on deity the AI building costs are reduced 40%. A modern era AI however will have an additional 25% reduction, (modern era = 5 times the -5 PerEraModifier), making it an impressive 55% reduction.

[nbarclay]

Setting a goal of human parity with the average AI would create a situation where human players typically lose a lot more games than they win, at least if they stick to a mostly builder playstyle. In order to have a 50-50 chance of winning any given game, a human player needs parity with the leading AI or AIs. Of course some players prefer to play on difficulty levels where they can win a lot more reliably than that, while others like challenging themselves trying to beat levels where they usually won't win. But for games to be reasonably competitive, they can't be too far away from parity between the player and the leading AIs.

[V3nom]

Originally posted by nbarclay
Setting a goal of human parity with the average AI would create a situation where human players typically lose a lot more games than they win, at least if they stick to a mostly builder playstyle. In order to have a 50-50 chance of winning any given game, a human player needs parity with the leading AI or AIs. Of course some players prefer to play on difficulty levels where they can win a lot more reliably than that, while others like challenging themselves trying to beat levels where they usually won't win. But for games to be reasonably competitive, they can't be too far away from parity between the player and the leading AIs.

You bring up a good point.

perhaps use a weighted average that is more dependent on the most powerful AI or two than the less powerful?

actually this makes alot more sense. Otherwise the player would be able to exploit mechanics by constantly gifting up extremely weak AI to just below the level they'd become a threat in order to lessen the bonus given to the most powerful AI.

[Nacht]

Originally posted by nbarclay
The ideal would probably be to have discounts come into play not just when other civs research a tech but also when other civs research successor techs - either techs that the tech in question is a prerequisite for or techs that are a lot later in the tech tree. Then the more primitive a tech becomes compared with the state of the art and with the general state of the world's technology, the bigger the discount would be. But setting up such a mechanism would be a whole lot more complex than editing XML files, assuming there's a way the modding community can do it at all.

It would be more complex but not impossible.

Since I play mosty with tech trading off I wanted to change this myself.

The function calculateResearchModifier in CvPlayer.cpp determines the research costs. It only uses two external values: the known tech modifier and the number of prereqs modifier, so XML tweaking will do no good.

A few things I though of:

Check the associated era of the tech. Loop through all players and give a reduction based upon the number of players already in later eras. (and if some are already 2 eras ahead make it ultra cheap )

Loop all players, check their total amount of techs, compare with total amount of techs of the current player and give a reduction if the gap is too big.

More advanced mechanics like the suggested checking of successor techs are certainly possible. (but wouldn't effect techs like horseback riding and divine right obviously).

Since I was already thinking about modding this function I would gladly provide a modded dll if we can come to agreement about the details

[nbarclay]

How hard would it be (if it's practical at all) to do a loop that calculates the total value of all of the techs that each civ has? That would keep players' skipping some of the early techs from skewing the comparison.

I'm skeptical of how well calculations based on differences in era would work. If I remember correctly, a CS slingshot puts a player in the third era even though he may still be missing several first-era techs. And single-era differences can very easily be a result of a civ's having researched "deep" rather than "wide," not a result of a genuine difference in overall technology.