[IDEA]Scientific Revolutions

[void_genesis]

Is the main issue that a human player with a tech advantage is unstoppable by any number of warring trailing AI civs (partly due to the carpet of doom limitations).

I guess what bothers me about civ is it's lack of historicity. All real world civs eventually collapsed under internal and external pressures. The current game only has happiness to slow expansion, and conquest to potentially lose territory. Apart from that it is just upward and onward. Real history (and current events) demonstrate how a technologically backwards nation can effectively engage with advanced ones if they use appropriate tactics.

Ideally for me achieving success in the game (dominance in any measure that leads to a VC) should be dangerous, in a way that gives the player new challenges to manage and new decisions to make.

 

[Ahriman]

I guess what bothers me about civ is it's lack of historicity. All real world civs eventually collapsed under internal and external pressures. The current game only has happiness to slow expansion, and conquest to potentially lose territory. Apart from that it is just upward and onward.

This has been long discussed in the guise of Civ and empire building games. The main problem is; decline and fall usually isn't fun. A whole part of the appeal of these games is building stuff. Having it torn down "just because" really isn't much fun for the player.
That's the difference between a game and a history simulator, I guess.
Also: *all* civilizations collapse? How about Japan?

 

[tlaurila]

That's the difference between a game and a history simulator, I guess.

Exactly.

As for the "alliances", I think you guys are thinking much more complex possibilities than I had in mind. With such alliances, yes, results are easily varied and unpredictable, as Txurce said, you don't even know whether you're making the end game harder or easier. Instead of alliance, think in terms of protectorate: A weak civ gives to stronger, in exchange for protection, making strong stronger.

I was simply thinking having a weak civ (civ W) capitulate and go "**** this, I'm giving up, and I'm donating all my cities and units to ... those guys (civ S)!". And then give over all units, and make S "conquer" all those cities. You made a likely strong Civ stronger, and more challenging to player.

One potentially interesting additional point is that civ W could come back by the Liberation mechanism, looking that those guys couldn't protect them anyway, and they'd be better off surviving on their own after all (at which point in an ideal world their AI would change to a Swiss-like less aggressive survivalist)

 

[Txurce]

I was simply thinking having a weak civ (civ W) capitulate and go "**** this, I'm giving up, and I'm donating all my cities and units to ... those guys (civ S)!". And then give over all units, and make S "conquer" all those cities. You made a likely strong Civ stronger, and more challenging to player.

One potentially interesting additional point is that civ W could come back by the Liberation mechanism, looking that those guys couldn't protect them anyway, and they'd be better off surviving on their own after all (at which point in an ideal world their AI would change to a Swiss-like less aggressive survivalist)

That's a wild idea. Presumably it would need safeguards to prevent a viable small cultural civ from merging, and something to keep civs from merging with the human player. I'd guess this is beyond the scope of the possible, but if not, would be really interesting to test.

 

[tlaurila]

That's a wild idea. Presumably it would need safeguards to prevent a viable small cultural civ from merging, and something to keep civs from merging with the human player. I'd guess this is beyond the scope of the possible, but if not, would be really interesting to test.

Btw, years back playing FreeCiv on a LAN, people giving up and donating their assets to another player is precisely what was happening to lead to 2 superpower endgame matchup

Technically, the idea should be relatively simply. Deciding on balanced factors of when to capitulate, and for whom's benefit, would be another matter, though.

 

[Sneaks]

I am extremely iffy on this idea . There are too many ways to inflate score in very quick manners, which could be manipulated by the player to receive free annexations.

 

[Txurce]

I am extremely iffy on this idea . There are too many ways to inflate score in very quick manners, which could be manipulated by the player to receive free annexations.

Yes, my first assumption was that this would have to be coded out.

 

[Ahriman]

I am extremely iffy on this idea . There are too many ways to inflate score in very quick manners, which could be manipulated by the player to receive free annexations.

Yes, I think the risks in a major change to how the game plays from a mechanic change like this are larger than the potential upside. So I wouldn't recommend going down this road.

I would prefer to make it easier for AI A to conquer and take over AI B and for this to make A more powerful than to design some kind of system where A and B can semi-merge.

 

[Sneaks]

I think in general we are focusing on the wrong end of the spectrum here and the wrong approach.

The current annexation idea is using two design concepts I don't necessarily agree with:
1. Late game can be best improved by limiting the number of remaining civs
2. Any late game bonuses should be applicable only to the AI

 

[Ahriman]

The current annexation idea is using two design concepts I don't necessarily agree with:
1. Late game can be best improved by limiting the number of remaining civs
2. Any late game bonuses should be applicable only to the AI

I don't think it is necessary to limit the number or remaining civs, but I do think it is necessary to allow the development of a superpower AI player.

I generally dislike bonuses that apply only to the AI. Sometimes they are necessary, but they should be of the form of invisible economy boosts, rather than mechanics that only the AI can benefit from.

 

[tlaurila]

I would prefer to make it easier for AI A to conquer and take over AI B and for this to make A more powerful than to design some kind of system where A and B can semi-merge.

Yes, this would be preferable in large part because it's an implicit mechanism to making the game harder, nothing new needs to be glued on top. Ironically, probably the biggest hindrance to this are the several VCs introduced with the very idea of making the end of game more interesting. Conquering would naturally only happen with militaristic civs, and a game where your strongest AI opponent is always militaristic grows stale, too.

 

[Ahriman]

the biggest hindrance to this are the several VCs introduced

A large/powerful civ can be a rich civ (for diplomatic victory) or a scientific civ, and couple potentially win either.

One thing that might need looking at is what drives the AI's annex vs puppet decisions. I'm guessing that the AI doesn't understand how much weaker puppets are now. Does it know enough to annex much more, unless it is going for cultural victory?

Conquering would naturally only happen with militaristic civs, and a game where your strongest AI opponent is always militaristic grows stale, too.

I also don't think that large AI empires that have been successful in war/conquest are necessarily the militaristic ones. Militaristic civs will declare war more often, but that doesn't mean they'll win wars more often. A peaceful civ might well tech faster and have better units and be able to field more of them later on with a better economy and so be more successful in war. Very often I have seen an aggresive civ attack a neighbor, get beaten back and then stagnate when someone else comes though; in a recent game I have, France attacked Songhai and then me and didn't manage to capture anythnig, and so when Siam stomped through with its massive horde of elephants it was thoroughly trounced. So at the moment, Siam is the potential leader, precisely because they avoided early wars and teched, and then brought in their elephants once the warmongers had exhausted themselves against each other. It feels very historic, like how in the real world Arabians managed to forge a massive caliphate after the Byzantines and Persians had utterly exhausted themselves (and their armies and economies) in centuries of fruitless war.

 

[Txurce]

I also don't think that large AI empires that have been successful in war/conquest are necessarily the militaristic ones. Militaristic civs will declare war more often, but that doesn't mean they'll win wars more often. A peaceful civ might well tech faster and have better units and be able to field more of them later on with a better economy and so be more successful in war. Very often I have seen an aggresive civ attack a neighbor, get beaten back and then stagnate when someone else comes though; in a recent game I have, France attacked Songhai and then me and didn't manage to capture anythnig, and so when Siam stomped through with its massive horde of elephants it was thoroughly trounced. So at the moment, Siam is the potential leader, precisely because they avoided early wars and teched, and then brought in their elephants once the warmongers had exhausted themselves against each other. It feels very historic, like how in the real world Arabians managed to forge a massive caliphate after the Byzantines and Persians had utterly exhausted themselves (and their armies and economies) in centuries of fruitless war.

This is without a doubt the way the game tends to play out. Exceptions are when a Germany, Greece or Rome win their first couple decisively. For some reason, the Aztecs, Danes and especially the Mongols eventually fall by the wayside, even if they do have big empires midway. Japan tends to fall just short of the first group in my games.

 

[void_genesis]

One consideration with the existence of multiple VCs with corresponding playstyles is to consider the fun value when the two leading civs are approaching their respective VCs.

So
A: conquest versus conquest is fun (lots of strategy and excitement) as long as the leading AI manages to keep up and develop into a super-power.
B: conquest versus peaceful science/culture/diplo is fun (asymmetrical offense vs defense)(unless the AI is unable or unwilling to attack you close to completing a VC)
C: peaceful versus peaceful is often boring (unless the AI happens to be neck and neck with you).

Molding the game toward each of these end-games could be fine tuned if there was a process that selected the civ closest to a VC part-way through the game, then gave it enough bonuses to become a super-power. The current score system is inadequate since it doesn't recognize tall empires. Maybe a better scoring system could be modded from info-addict.

 

[Thalassicus]

I feel there's two main things hindering AI conquest:

• Firaxis changed the open terrain modifier to -10%, which favors attackers less than the -20% I had in the mod (original vanilla value was -33%).
• Firaxis blocked AIs from attacking citystates in the early game.

These two things had a big impact on the capability for militaristic AIs to do well. I've been wanting to change #1 for a while... just hadn't seen much feedback indicating it might be necessary. I could add a change to the terrain modifier in the next beta?

 

[Ahriman]

Firaxis changed the open terrain modifier to -10%, which favors attackers less than the -20% I had in the mod (original vanilla value was -33%).

I disagree with this. A larger open terrain penalty favors defenders, a smaller penalty favors attackers. Attackers are typically the ones who have to go out onto the more open tiles, and are the ones getting shot at more.

Do I don't think a change is necessary. Larger penalties also end up punishing the AI much more relative to the player, since the AI is worse at placement.

 

[Thalassicus]

I did extensive testing with it half a year ago. Out of 50 autoplay games of 200 turns apiece, higher attacker modifiers uniformly resulted in more superpower AIs. This variable has the most effect on superpower formation out of anything I've experimented with. Player discussions at the time supported this conclusion in non-autoplay games as well.

(The tuner has an autoplay feature where Civ plays all the leaders as AIs.)

I didn't keep the mod's original 20% modifier because I wanted to remain close to vanilla, where they dropped from 30% to 10%. I don't feel it was a good decision by the developers to change it so much.

 

[Ahriman]

Out of 50 autoplay games of 200 turns apiece, higher attacker modifiers uniformly resulted in more superpower AIs.

Fascinating. I find that odd, I wonder if the larger penalty ends up favoring the side with the larger army, typically the attacker. Increasing the penalty would be a boon to the human though.

Another related thing I've wondered about; is there any way to have the open terrain bonus for ranged units *not* apply when shooting at naval units? I find this frustrating, it makes archers and crossbows too effective vs ships, and I don't see why open terrain archers should be better vs naval than rough terrain.

 

[Txurce]

I feel there's two main things hindering AI conquest:

• Firaxis changed the open terrain modifier to -10%, which favors attackers less than the -20% I had in the mod (original vanilla value was -33%).
• Firaxis blocked AIs from attacking citystates in the early game.

These two things had a big impact on the capability for militaristic AIs to do well. I've been wanting to change #1 for a while... just hadn't seen much feedback indicating it might be necessary. I could add a change to the terrain modifier in the next beta?

I disliked the Firaxis change because it reduced the value of positioning, but like Ahriman, figured that it would ultimately hurt the AI vis a vis the human player. Viewed from the context of this thread, I'd be in favor of reverting it to 20% or higher, as you originally had it.

And speaking about ways to create stronger AI for the universally acknowledged weakness of the late game, what do you think about my suggestion (encapsulated in post #67)?

 

[void_genesis]

I did extensive testing with it half a year ago. Out of 50 autoplay games of 200 turns apiece, higher attacker modifiers uniformly resulted in more superpower AIs. This variable has the most effect on superpower formation out of anything I've experimented with. Player discussions at the time supported this conclusion in non-autoplay games as well.

(The tuner has an autoplay feature where Civ plays all the leaders as AIs.)

I didn't keep the mod's original 20% modifier because I wanted to remain close to vanilla, where they dropped from 30% to 10%. I don't feel it was a good decision by the developers to change it so much.

Trust Thal to find a simple and elegant solution to a superficially complex problem. I had forgotten that earlier versions of ciV would more often than not give you a hefty run away ai to pull apart in the end-game. Give me an average 25% chance of losing the end-game after starting out strong and I will be very happy!