A Better AI.

[Roland Johansen]

Here are the justifications for FLAT bonuses across the board.

AI barb/animal bonuses: The AI is tactically weak which means it suffers more at the hands of barbs, this tactical weakness is applicable at every level.

AI upgrade bonuses: The AI makes relatively poor use of it's upgrade cash relative to what even the dumbest human would - even a not so smart human will instinctively upgrade units near hostile borders. The AI, even improved, does not have those instincts - this is true on every difficulty level. The AI also relies more on "Quantity over quality" when it comes to winning wars and making them pay full price would harm them more than it does a human who targets upgrades much more smartly.

AI supply bonuses: Again "Quantity over Quantity". If the AI is to have a prayer of succeeding, it needs to send in more units than a human would, and/or the units spend more time milling around rather than getting to the point of the invasion. Hence a discount is justified.

AI war weariness bonuses: Again Q/Q is a factor, but also the AI is intrinsically worse at manipulating this game mechanic, also at higher difficulties the AI relies all the more on quantity while humans are all the more better at manipulation hence a scaling on difficulty is justified.

AI inflation bonus: Ok I don't have a good reason for this one, other than as an extra free pass on top of other commerce free passes. Regardless, since inflation is already reduced by other difficulty-sensitive bonuses there's no need to scale inflation by difficulty too (in particular there are the per-era bonuses which means AI expenses go down over time.

I can see why you would like to give the AI some bonusses at the lower levels of difficulty and the AI will benefit from those bonusses as you describe. But at the same moment you're decreasing the AI bonusses at higher levels of difficulty.

I have already described in many words what problems this could cause. It's not very usefull to do that again, but I'll try to put it in one paragraph which might make any discussion between the posters in this thread somewhat more structured.


At the higher difficulty levels the AI has a bigger production bonus and thus more units. More units leads to higher upgrade costs, higher supply costs in foreign lands when at war and more war weariness when attacking an enemy. Without an adjustment to these factors that is proportional to the increase in number of units at the higher difficulty levels, the AI will function like a machine with some parts moving faster than others. (my personal view on this subject)


By the way: good to see you here, Zombie. You must love the way that the AI can now pop-rush as you wrote an article about pop-rushing (and other micro management).

 

[icemanjsg]

When is the update which gets rid of the 15 seconds turn load time coming out, I read it was coming out late last night or early this moring did i miss it?

 

[Zombie69]

By the way: good to see you here, Zombie. You must love the way that the AI can now pop-rush as you wrote an article about pop-rushing (and other micro management).

Nice to see you're still here. Yeah, i liked those improvements. However, i was a bit distraught to find in game that some Warlords stuff had been implemented in this dll, like the pop rush bug being removed. I didn't expect the dll to change this aspect in Vanilla civ.

Also, i don't like any of the changes to the governor AI (the way it worked before was better for my taste). Also, i tend to find that even though the AI now makes more units, he seems even stupider than before about how to use them.

 

[Roland Johansen]

Nice to see you're still here. Yeah, i liked those improvements. However, i was a bit distraught to find in game that some Warlords stuff had been implemented in this dll, like the pop rush bug being removed. I didn't expect the dll to change this aspect in Vanilla civ.

Also, i don't like any of the changes to the governor AI (the way it worked before was better for my taste). Also, i tend to find that even though the AI now makes more units, he seems even stupider than before about how to use them.

I see what you mean. Not all changes to the AI work for the best or at least not immediately. There are some changes that need to be refined more. I also sometimes have problems with the governor. I'm not exactly sure if it is just because I'm more used to the old governor. The emphasize effects are different now. I've posted a few screenshots here in this thread where the governor was using a clearly suboptimal selection of tiles, so I guess that Blake and Iustus know that there are some problems with the governor.

If you can pinpoint some inefficiencies in the AI, then you should post them here. There are just two guys working on this and they can't find every mistake that the AI makes. You know how to play this game at the highest levels so at least for me, your opinion counts.

 

[Arlborn]

By the way people, this changes in the handicap were more about what HE thought was better for the whole scene. This deffenetelly DONT need to go togheter with the DLL if you dont want to(as far as I know). And also, you can really easily change it yourself.

 

[Iustus]

(Apologies, in advance, if this question has already been answered in the over 70 pages of this thread.)
I can't seem to verify that I installed "BetterAI" correctly. I followed the first procedure in the thread quoted, but when I mouse over score, the words "Better AI" do not show up.
Any suggestions on what I might be doing wrong?
Thanks so much.

(Installing it as a mod does seem to work, though.)

You do have the alt key down right? Look carefully at the path which you have the DLL at and make sure it is identical to the one I have pictured. I suspect you have it burried one level too deep. If it is inside CustomAssets, then it should not be inside a folder called BetterAI.

-Iustus

 

[Tiberias]

At the higher difficulty levels the AI has a bigger production bonus and thus more units. More units leads to higher upgrade costs, higher supply costs in foreign lands when at war and more war weariness when attacking an enemy. Without an adjustment to these factors that is proportional to the increase in number of units at the higher difficulty levels, the AI will function like a machine with some parts moving faster than others. (my personal view on this subject)

Depends on how much the computer player's income scales with difficulty. If the computer player's income is going up much faster with higher difficulty levels than the number of units they produce, it's not a problem. If, on the other hand, their income isn't scaling as quickly (or not at all), than you're right, it's a serious problem; the AI will quickly run out of money due to maintenance and upgrade costs.

The War Weariness point is a good one, and I suspect that the higher number of units will in fact increase War Weariness for an offense-heavy computer player more quickly at high difficulties. Some scaling may be needed there.

 

[Wodan]

Strange that people don't like the changes to the governor AI. (By that, I assume you mean the way it allocates citizens to work which tiles and/or be specialists.) Personally I love the way it works now. Much improved over both previous versions as well as the game default.

Wodan

 

[Iustus]

New Build 1/8, we are now feature complete and pushing toward our 1.0 release.

War and tactics:
- Increased the number of defenders trained and used in interior cities.
- Increased the weighting on needed defenders in the presence of hostile units.
- Improved the AI's assessment of whether or not it's safe to leave a city to kill interlopers.
- Decreased the chance of Spearmen being trained.
- Drill promotion line now synergizes strongly (protective should pursue it sometimes).
- Pillagers may join city attacks in progress.
- Changed some things with stack of doom formation and breakup.
- The AI should be more flexible in declaring wars.
- The AI should be more aggressive in offensive wars, especially on Aggressive AI setting (try it!).
- "We have enough on our hands" is now used as a no-war excuse of last resort.

Governor and Workers:
- Fixed the governor bug with expansive workers.
- Wrote new cottage-valuing logic for governor (town > village > hamlet > cottage).
- AI workers should put more weight on production, especially in the later game.
- Certain AI's will put more weight on production according to personality.

Cultural Victory:
- AI's should be slightly less likely to pursue cultural victories.
- Cultural AI's use better logic for deciding when to turn off research.
- Improved Cultural AI war-making habits.

Improved Explorer Behavior (Automated and AI)

Issues:
- Lag issue resolved.
- Infinite loop issue resolved.
- Fixed the crash bug with moving units into a friendly city.

-Iustus

 

[icemanjsg]

AWSOME AWSOME AWSOME

best mod for civ 4 ever

 

[Tiberias]

Very nice, thanks much. I'll give this a try tonight . . . maybe I can go back to my Huge/Earth1000AD game and actually be able to play it .

 

[Blake]

At the higher difficulty levels the AI has a bigger production bonus and thus more units. More units leads to higher upgrade costs, higher supply costs in foreign lands when at war and more war weariness when attacking an enemy. Without an adjustment to these factors that is proportional to the increase in number of units at the higher difficulty levels, the AI will function like a machine with some parts moving faster than others. (my personal view on this subject)

I disagree. At higher levels the AI has a stronger economy because it grows faster builds infrastructure faster techs faster (etc), the AI is VERY good at adpating to bad happiness and stuff now, in fact it should pretty much always focus on whatever aspect of it's economy is weakest, this means it can adapt to added expenses or added unhappiness in a way it once couldn't (or at least not effectively). Making the AI feel some pressure in certain areas can also help to direct it's development - rather than playing in a total economic sandbox - the AI is now adaptive enough that it doesn't NEED the sandbox.

While this does mean the AI can no longer flash-upgrade it's entire army the AI doesn't have a problem with finding "other" ways to recycle it's units such as mass zerg attacks on cities. The AI has no compulsion to raise cash until all of it's units are upgraded, it tries to upgrade some with petty cash and will mass upgrade after a cash boon (a cash bomb will be worth ~30-50 units) but it wont go to 100%g to upgrade units and thus it's economy wont be crippled by the presence of a mass of obsolete units. The only difference is that it'll have more obsolete units around for longer, which is basically what people want - no instant upgrade of the ENTIRE army.

By the way people, this changes in the handicap were more about what HE thought was better for the whole scene. This deffenetelly DONT need to go togheter with the DLL if you dont want to(as far as I know). And also, you can really easily change it yourself.

Thank you for saying that .


I'm also going to quote someone from another thread:

When the AI mass upgraded, I actually wanted to find Blake and throw him out the window too for making the AI so much better at production while leaving its upgrade cheats in. This is not fun, or balanced.

I do live on the first floor of a building so the fall would not be harmful but I recognized uberfish's point as being a very valid one.

Can't make everyone happy can only use my judgment to the best of my ability.

 

[icemanjsg]

BTW I can't see huts graphics

also I can see the blue circles with scouts in the shroud ??

 

[Harrier]

I'm also going to quote someone from another thread:

When the AI mass upgraded, I actually wanted to find Blake and throw him out the window too for making the AI so much better at production while leaving its upgrade cheats in. This is not fun, or balanced.

I do live on the first floor of a building so the fall would not be harmful but I recognized uberfish's point as being a very valid one.

Can't make everyone happy can only use my judgment to the best of my ability.[/QUOTE]

I'm with you on this one.


Edit.

Your judgement that is, not you being thrown through the window.

 

[peacenik]

Blake/Iustus---Thanks again for this great improvement to Civ4. I hope release 1.0 won't be the end of your project.

 

[Roland Johansen]

Depends on how much the computer player's income scales with difficulty. If the computer player's income is going up much faster with higher difficulty levels than the number of units they produce, it's not a problem. If, on the other hand, their income isn't scaling as quickly (or not at all), than you're right, it's a serious problem; the AI will quickly run out of money due to maintenance and upgrade costs.

The War Weariness point is a good one, and I suspect that the higher number of units will in fact increase War Weariness for an offense-heavy computer player more quickly at high difficulties. Some scaling may be needed there.

I guess that you mean supply costs instead of maintenance costs. Maintenance costs suggest unit upkeep and Blake isn't touching that (at the moment). I just want to make it very clear what we are talking about.

To compare the income of the AI at various difficulty levels, you should not compare the income at the same year. For the purpose of upgrading its units from archers to longbowmen, it isn't interesting how well developed the economy is in 200BC or 500AD at the various difficulty levels. It is interesting how well developed the economy is when the AI develops/invents feudalism. My guess is that when you compare the level of development at this moments in the game that there is no big difference in level of development between a deity level AI or a noble level AI. It's just that the deity level AI reaches that moment sooner. The biggest difference between a deity level AI and a noble level AI's economy at the moment they discover feudalism is probably that the deity level AI has lower city and civic upkeep cost. But since that is only a part of the economy of the AI, it will not cover the 3 times as many units that need to be upgraded.

But in the end you're right. It fully depends on how much the AI's income is going up with difficulty level. As long as you compare them at the moments that certain technologies are developed and not fixed time points. And I don't think their economy scales enough to keep up with the increased number of units that have to be upgraded.

Glad to see that we agree on the war weariness issue.

New Build 1/8, we are now feature complete and pushing toward our 1.0 release.

-Iustus

This sounds like a great update. Especially the war AI and governor changes sound very good. I'm going to continue my game with this new version. Thank you for a better civ experience.

 

[Blake]

To compare the income of the AI at various difficulty levels, you should not compare the income at the same year. For the purpose of upgrading its units from archers to longbowmen, it isn't interesting how well developed the economy is in 200BC or 500AD at the various difficulty levels. It is interesting how well developed the economy is when the AI develops/invents feudalism. My guess is that when you compare the level of development at this moments in the game that there is no big difference in level of development between a deity level AI or a noble level AI. It's just that the deity level AI reaches that moment sooner. The biggest difference between a deity level AI and a noble level AI's economy at the moment they discover feudalism is probably that the deity level AI has lower city and civic upkeep cost. But since that is only a part of the economy of the AI, it will not cover the 3 times as many units that need to be upgraded.

I think you're wrong about the AI's level of development. The deity level AI can quite trivially build infrastructure since it gets huge discounts on non-wonder buildings and the AI does like building cheap stuff.

I also think you're wrong with this idea of units that "need to be upgraded" - not every unit needs to be upgraded, humans don't upgrade every unit (usually "Recycling" the poorer ones) and the AI only (used to) upgrade every unit because it can trivially afford to. It might be that it needs to be made less aggressive in upgrading units (ie neglecting those with low exp and low strength) but if so that's a problem with the AI management not with the new handicaps scheme.

 

[kettyo]

Big Big thanks boys !!!

You should be included in the game credits list.
I'll see if i could do that

 

[The M'Hael]

Thanks for the new update. (especially the lag fix )!

As for the bonuses, I trust that Blake did what he thought would be the most rounded for game play.

In on of my games, I attack HC with a stack of ~20 units, 5-6 of them siege weapons. I lost sadly killing only 2 units. I later, about 20 turns, bribed Shaka into attacking HC. He marched in about the same # of units that I had. HC even put more units into that city, yet Shaka managed to take the city and vassal HC. What's up? Note that this is only on Warlords difficulty with 1/1. I didn't know that AI's got that much of a bonus against other AI's...

Any explanation?

Will try with 1/8 and see if anything changes...

The M'Hael

Edit: Oh by the way, you guys should check the BetterAI forums. Someone said something about inlining small functions. He said it could be done to increase the processing power of the AI... He posted there a few days back. Since no one else has posted, figured you hadn't seen it...

 

[kettyo]

Strange that people don't like the changes to the governor AI. (By that, I assume you mean the way it allocates citizens to work which tiles and/or be specialists.) Personally I love the way it works now. Much improved over both previous versions as well as the game default.
Wodan

I like the homogene specialists approach and the slavery overpopulation.

I last played some december build and there the governor might valued hammers a bit too low for my taste but it could have changed since.

Overall it's surely an improvement over the old one.

I haven't tried the new build governor yet but i shall.