A Better AI.

[Wodan]

Yeah but only 1 catapult.
I think it could need a lookafter.
They should include many siege weapons in the dagger-army otherwise they could lose a lot against city garrisons.

IMO a "dagger" army does not include siege weapons because "dagger" (by definition) is before Construction.

I think Hatty's huge army is the leftovers of a dagger choice without pulling the trigger.

Wodan

 

[AKS1911]

Playing on monarch/marathon I have some weird feelings about the pace of the game with the improved AI, wich leads me to the following questions :

- human diplomacy malus
The AIs are stronger, yet they still seem to only focus on me and my friends even with juicy targets next to them. What I mean is that I never see two civs who don't like me (to the point that I cannot bribe neither of them) and hate themselves too, engage in their own war, not related in anyway to me or my friendly AIs. It seems that if the AI has me/my friends and another AI as potential targets, it will target me/my friends even if it is much less logical/rewarding.

-> Question is : do those maluses on the human translate into maluses for how the AI chooses it's targets?

- 0 gold for trade
With 18 civs, the typical thing is to see 10 civs with 0 gold, 5 with very small amounts and two with decent amounts.
This means I'll get raped even more on many tech deals, as the AI won't be able to add some gold to even the already biaised trade. Of course 100 gold traded can equal 200 research beakers, so this translates into another malus put on the human, not because of game/difficulty settings, but because of how the AI manage their economy.

-> Question is : Does the inability to add gold in trades hurt some AIs too between themselves, as without cash, they very well may miss many deals?

- AI troops upgrade
They have HUGE discounts on unit upgrade costs.
Exemple, monarch/marathon : on monarch they pay only 25%, correct?
Small map, upgrade cost = 380. AI saves 285. 10 units, 2850 gold.
Huge map, upgrade cost = 380, AI still saves 285. 100 units : 28 500 gold.
Rifles tech on small map : 8280. On huge map : 12420.
Remember that gold spent paying maintenance may equal twice in beakers because of buildings/national wonders?

Basically, we can say that when an AI reaches rifles on small/monarch/marathon, it roughtly freely gains half a tech, but on huge/monarch/marathon it gains about 4.5 free techs? (I hope I missed something somewhere, this is nonsense!)
If they are balanced with this bonus in mind, then they are too weak before any troop upgrade takes place. Early war all the way :-/

-> Question is : Am I the only one here thinking that an AI economy designed with such bonuses results in MALUSES for it before they kick in?

- AI research frenzy
I think that they spend too much on research. i don't mean "oh they research much faster than me, I want my candy". No.

I mean that :
* given their efficiency at using their military, they have too weak armies. Crush their initial attack strenght and conquer them. Not enought reserves.

* it is SAD to see lvl 1 mech infantry. They just upgraded that lousy 4000 bc archer, again and again, never replacing it with a unit with better xp.

* their too weak reserves leave them vulnerable to "second fronts" openings. Under attack? Bring a friend and enjoy. The only problem is the inital attack forces, after, they just suck at rebuilding forces.

* No gold = no help. You can bribe AIs into your wars with gold too, you know? If a technology backward civ is under attack, it just CANNOT bribe another AI into helping him if it has no tech to give it. Remember that this is exactly what vassal states do! "You get half my income if you help me". But vassal states are much less flexible (too far, too many ennemies at once for the master to take so he refuses, etc...)

* No gold but some techs? Well, it would had cost the AI 2500 gold to bribe this other AI to come to it's help, but, no, it will be a nice 12000 beakers tech, thank you...

* Even with their huge upgrade bonuses, I've often found myself killing obsolete units, even when the AI had the proper tech, because... It just spent too much on research and had no cash reserve to mass upgrade when my army strike!

* No gold = no more rush buying when the whole world toggles into emancipation?

* No gold per turn means no >purchase< of ressources. Yay! Beeline to genetic to get those + health, AI friend...

I think that the improved AI economy just reveals how badly the AIs use their income.


-> Question is : can you modify the AIs budget spending so that it stops being so obsessed with research that it shoots itself in the feet?

Hummm... Maybe AFTER military AI upgrades

Anayway, I can't wait for the next build

 

[Instant_Cereal]

Finally somebody else who sees and agrees there's something weird with the pace of the game, especially on marathon games.

 

[drkodos]

Malus? What is this?

I do not think you mean Apple trees.


Marathon was skewed even before the game engine improvements. It lenghthens the time for research disproportianate to constructions times for units and buildings.

Marathon makes the game easier, across the board.

 

[Iustus]

New download 11/22 should fix the gold problems and a lot more.

(Sorry for the brevity, its pre-Turkey day!)

-Iustus

 

[kniteowl]

Changes 2006-11-22

- AIs should build wonders more frequently
- tactical changes, AIs should stack attack properly now
- AIs more likely to build barracks, and give other AIs more military respect
- AIs should attack with more focus
- AIs in financial trouble are much more likely to pick buildings and techs to solve it
- AI missionary strategy improved
- bug where AIs would run 90% research and run out of gold always fixed
- holding down ALT when mousing over the active player will show "=== Better AI ==="
- AIs will emphasize the right things for city plots
- minor bugfix with healing units rejoining groups early

Copied the "Released Note.txt" so everyone knows waht the new Improved AI patch does.

 

[Iustus]

two late addtions to the release notes:
- possible crash with stack attack fixed (when every unit is very damaged)
- crash with air units fixed

-Iustus

 

[kettyo]

Malus? What is this?

I do not think you mean Apple trees.


Marathon was skewed even before the game engine improvements. It lenghthens the time for research disproportianate to constructions times for units and buildings.

Marathon makes the game easier, across the board.

I didn't know that (i play marathon very rare it's so long).

So research or production is lenghtened more relatively?

 

[kettyo]

Malus? What is this?

It's just another word for handicap.
AFAIK for instance insurance companies use this term while 'handicap' is mostly used in sport games.

 

[Thedrin]

- 0 gold for trade
With 18 civs, the typical thing is to see 10 civs with 0 gold, 5 with very small amounts and two with decent amounts.
This means I'll get raped even more on many tech deals, as the AI won't be able to add some gold to even the already biaised trade. Of course 100 gold traded can equal 200 research beakers, so this translates into another malus put on the human, not because of game/difficulty settings, but because of how the AI manage their economy.

-> Question is : Does the inability to add gold in trades hurt some AIs too between themselves, as without cash, they very well may miss many deals?

Look back through the list of improvements made. The amount of gold listed in the trade screen is not the amount of gold the AI has, just the amount that the AI is willing to trade (this was always the case). One of the changes was to get the AI to hold back on more of their gold so that the AI could use this gold to fund a higher researching rate.

Answer: No.

 

[Arlborn]

Arlborn,

A Monty/Shaka/Louis vs Hatsy/Ghandi war seems imminent in your current game

Neither side could be called your friends really.
No matter i'd join Hatsy in your place and try to hit Louis's culture citadels.

In fact, Shaka and Monty did atack Kublai Khan toghter, but hey, Kublai khan also had a big army! He did raze a city of Monty and a city of Shaka just under their nooses!

Why the heck almost every single AI(exclude Gandhi and Asoka) has a HUGE army now? Way bigger than their normal before your mod??? They all plan daggers atack in the Industrial era or what?? Hathy and Louis also have a huge army but didnt go to any war yet! Is it how it supposed to be now? 0.o I had to build much more army than I usually build and it is still really weak compared to the others(beside the Indians)! I'm scared, already paying some unit cost and still not big enough?

 

[skywalker]

They have big armies because they have fantastic economies which give them enough production to build those armies.

 

[Zarah Neander]

-> Question is : Does the inability to add gold in trades hurt some AIs too between themselves, as without cash, they very well may miss many deals?

Probably not. Turning on debug mode and watching the AI deals is rather .. interesting. Isabella gives CS to JC for 90 gp. JC was Hindu while Isa was jewish. Try to pull that of as a human. HC gives Fusion (yes, Fusion) to MM for Democracy. Seems inter-AI trades can go as low as 1:20 gold/beaker.

Not that I want to start another 'the AI gets too many bonuses' discussion. The AI needs every help it can get cause it has exactly zero long-term goals. And this isn't something which is trivial to implement. Most likely impossible.

 

[Harrier]

Marathon was skewed even before the game engine improvements. It lenghthens the time for research disproportianate to constructions times for units and buildings.

Marathon makes the game easier, across the board.

I think that was the point of Marathon - the slower research means you can fight wars against units of the same type/era. At standad speed more advanced units appear to quickly. So all though the war starts with Macemen/Knights it ends with Riflemen/Cavalry etc.

It does as you say make the game easier - most of my Domination victories have been achieved at Marathon.

 

[peacenik]

Many thanks to Blake and Iustus for the great improvements to Civ 4 and making them available to all of us. A nice holiday treat.

The recent builds seem to involve fine tuning and bug fixing. Are there any plans for more big changes to the AI??? The great changes so far make me hope for more.

Thanks again for this great work--the best 'mod' I've ever seen in the Civ series.

 

[player1 fanatic]

The recent builds seem to involve fine tuning and bug fixing. Are there any plans for more big changes to the AI??? The great changes so far make me hope for more.

Well, I think it's good thing.
That way you can make a stable version, which could be easily integrated into other mods, or used in some other purpose, with no chance of failure or some other problem.

Then, new version could be made, with new experimental solutions.

 

[peacenik]

Well, I think it's good thing.
That way you can make a stable version, which could be easily integrated into other mods, or used in some other purpose, with no chance of failure or some other problem.

Then, new version could be made, with new experimental solutions.

I agree completely---but my question is whether more major changes are going to be attempted. This could either be the finishing touches to a great mod, or the finishing touches to the first phase of a great mod. I'm hoping Blake will tell us which it is.

 

[Doomed_UK]

Yes - as much as I like all the changes I would quite like a definative stable version for now before the next set of improvements. (A stable version that firaxis may cast thier eye over before the next patch).

 

[Todd Hawks]

Yeah, I wouldn't mind a stable version either. I haven't really been able to play at all since the last patch since no version (neither the patch itself nor any of Blake's releases since) has worked without what I consider game-breaking bugs (be they crashes or infinite loops or simply game mechanics gone awfully wrong).

And since the patch is necessary for Warlords due to the vassal problems and since Blake's mod (with bugs) is integrated into the patch, I can't play anything at all. Well, maybe vanilla 1.61, but I'm not going to go there. I don't think it was a wise decision to implement a largely untested mod into the patch (and lets be honest, it wasn't tested. I remember how Alex was still "following Blake's development of the mod" for the patch a few days before the patch came out). If anyone wanted to use the mod, s/he could have just downloaded it... no need to make it a fixed part of the patch.

Not to diminish Blake's great work but somehow Civ IV has fallen apart with this decision with no stable version for me to play at the moment.

 

[Arlborn]

Yeah, I wouldn't mind a stable version either. I haven't really been able to play at all since the last patch since no version (neither the patch itself nor any of Blake's releases since) has worked without what I consider game-breaking bugs (be they crashes or infinite loops or simply game mechanics gone awfully wrong).

And since the patch is necessary for Warlords due to the vassal problems and since Blake's mod (with bugs) is integrated into the patch, I can't play anything at all. Well, maybe vanilla 1.61, but I'm not going to go there. I don't think it was a wise decision to implement a largely untested mod into the patch (and lets be honest, it wasn't tested. I remember how Alex was still "following Blake's development of the mod" for the patch a few days before the patch came out). If anyone wanted to use the mod, s/he could have just downloaded it... no need to make it a fixed part of the patch.

Not to diminish Blake's great work but somehow Civ IV has fallen apart with this decision with no stable version for me to play at the moment.

You are the first a hear to have this kind of bug with version 2.08 Without others patches of Blake..And I think his last version probably is pretty much stable.