A Better AI.

One thing I definitely notice in the latest build is the AI will whip defenders when I'm sitting at the gates. Which brings a new factor into conquest wars. When going up against a heavily defended city in the age of Slavery (and maybe the "draft" civic), you need to bring enough seige weapons to tackle the city defenses in a single round.
When possible you should always wait outside the cultural boundaries of your target city, then advance only when all your units are ready. I'm not sure exactly how it works, but the same city that has 6 longbows in it after you sit next to it waiting for all your troops to arrive may have only 1 or 2 to start.
 
12/21 build crashed: I tried to start a new game (huge Lakes with 18 civs), it crashes. If I make it 17 civs, it works. Back up to 18 and it crashes again.

...

oh and that other game I was playing (with an older build), the Genghis AI got to 62% land mass and 64% population (above the required amount). Arghhh! Only twwwoooo percentage and I would have seen the AI get its first dommination win. :(

It was close though. I was tempted to give him a city so he could win it.
 
Blake said:
(...) There's even a dilemma here when it comes to cheating. Take Terra map for instance, should the AI know that there is a second uninhabited continent out there in the black fog? It is actually pretty trivial to just "maphack" this information but the AI would be cheating, not in some cases since the human knows they selected Terra, but maybe in other cases like they chose Shuffle instead or it's a crafted scenario which has been designed as terra-ish but none of that info is given to the player - it's pretty hard (not sure if it's even possible) to tell what map script was used or if shuffle was used and in any case it wouldn't be right to hardcode behaviors to map scripts. The Human always has equal or better knowledge of the map - they either know what the map is, or they know they don't know. (...)

I agree that the AI knowing about the new world on terra could be cheating in some constellations.But I think this is a case where the AI cheating is actually healthy for game fun.The terra script is an extremly fun idea - competition about conquering the new world.A new world which is already inhabitet by strong natives (=barbarians).So it is not a cakewalk anyway...knowing about it is one thing, actually sending over enough units another.To make this fun in SP, the AI needs at least the knowlegde. Otherwise it is constant cheating for the human player (who will choose terra on purpose in most cases).
Of course abetter solution would be to give the AI a strong desire for discovery.If they have a coastal ciy, the should be eager to build ships to send them out - likely ending up with the ships discovering the new world on terra.
However, my feeling ATM is that the main problem isn't soo much the lacking knowlegde.Under my impression, the AIs learn about the continent quite fast - the problem is that they are for some reason very hesitating to start to explore and settle it.And if they do it at all, they are lacking usually enough forces to fight off the barbs.
 
12/21 build crashed: I tried to start a new game (huge Lakes with 18 civs), it crashes. If I make it 17 civs, it works. Back up to 18 and it crashes again.

I'm also experiencing this. It definitely doesn't approve of 18 civs.

EDIT: something else I noticed - in the past, when you told a unit to explore, if it was attacked and took damage, it would halt to heal and then continue exploring. Currently (not sure when this behavior started to occur), if an exploring unit takes damage, it continues to explore without healing. Is this deliberate?
 
Quote by Blake:
The most important question is the difficulty, AI's at prince and lower play kinda like human noobs in multiplayer, they have a tendency to go too light on defense.

Now im on the difficulty level somewhere between prince and monarch so im bouncing between. In the previous version the AI on prince was better with units. I dont know if what you ment by this that your AI is Dumber on prince and noble levels or its just a feature of the game. I think its better to keep the ai bonus that are in the game and worry about balancing those out later if they need it after the ai is in a good finished as can be build. But I dont want to see an AI that is "dumber" per different levels except maybe under noble, I play multiplayer with friends sometimes but they aren't as good. We choose different difficulties and with the average running on prince even they know that a city with 2 archers is just asking to be taken.
sorry for the rant. still love the new ai.
merry xmas :D keep up the good work, but not today :lol: jeez..
 
Running the 12/21 build here with 18 civs on huge/fractal with no issues (one isolated CTD the other day that hasn't shown up again). Running with 2GB RAM and a 256MB video card.

Folks who are having issues should probably indicate hardware specs?
 
Blake, I like your work but watch the posts above everybody is telling you the AI in the 12/21 build is broken. A monarch game has never been that easy. You can't say that the fact the AI doesn't build defenders or ancient wonders is a 'feature'... :rolleyes: I've always supported the project but I can't play if there is no challenge anymore... That's the same for all of us. Please fix these 2 bugs.

Merry christmas anyway
 
No code tags? Blake! I wanted to add this to my mod. WinMerge? I have more education in my future, I see...
 
Blake, I like your work but watch the posts above everybody is telling you the AI in the 12/21 build is broken. A monarch game has never been that easy. You can't say that the fact the AI doesn't build defenders or ancient wonders is a 'feature'... :rolleyes: I've always supported the project but I can't play if there is no challenge anymore... That's the same for all of us. Please fix these 2 bugs.

Merry christmas anyway


Blake, I just want to say thank you for this mod. I find that the computer keeps up in tech much better, and has much better production now. However, I felt it invaded a lot better prior to your mod and defended better as well. For example, yesterday as Louis I was in the middle of a world war with Qin and his two vassals Kublai Khan and Tokugawa with my two vassals Isabel and Wang holding the front lines. I shared borders on the other side with the Turks, who declared war on me. They invaded with a HUGE stack of about 20-30 units. As I had little to no army on that side of the world they should have crushed me. However, 95% of this stack was cats and trebs making it very easy for me to repel the invasion without loosing anything. This is a game I should have lost, or lost two/three cities. I hope you can fix this bug.

In any case thanks for all your hard work, and merry Christmas!!

OH, I was playing on Prince.
 
Blake, I like your work but watch the posts above everybody is telling you the AI in the 12/21 build is broken. A monarch game has never been that easy. You can't say that the fact the AI doesn't build defenders or ancient wonders is a 'feature'... :rolleyes: I've always supported the project but I can't play if there is no challenge anymore... That's the same for all of us. Please fix these 2 bugs.

Merry christmas anyway

I have to disagree - I always played at Emperor pangea big map 12 civ, with this mod on I play Monarch and it's just as hard - in different ways. I don't find the AI is so bad at wonders, nor is it any worse at war. It's in fact significantly improved in many ways, but we are starting to take these for granted in favour of pointing out little problems.
 
I have to disagree - I always played at Emperor pangea big map 12 civ, with this mod on I play Monarch and it's just as hard - in different ways. I don't find the AI is so bad at wonders, nor is it any worse at war. It's in fact significantly improved in many ways, but we are starting to take these for granted in favour of pointing out little problems.


Okay, so you disagree with the 3 other posters. What is your build ? Which difficulty level do you play on ? What kind of maps or options ? Thank you for the information.
 
I have to disagree - I always played at Emperor pangea big map 12 civ, with this mod on I play Monarch and it's just as hard - in different ways. I don't find the AI is so bad at wonders, nor is it any worse at war. It's in fact significantly improved in many ways, but we are starting to take these for granted in favour of pointing out little problems.


The problems (lack of defenders in most AI cities and few ancient wonders being built) are new and present only in the 12/21 (not in the 12/12 build I think) so if you didn't play in the last 3 days, that's why you didn't see it.
 
Blake, I like your work but watch the posts above everybody is telling you the AI in the 12/21 build is broken. A monarch game has never been that easy. You can't say that the fact the AI doesn't build defenders or ancient wonders is a 'feature'... :rolleyes: I've always supported the project but I can't play if there is no challenge anymore... That's the same for all of us. Please fix these 2 bugs.

Merry christmas anyway

Thanks for telling me I'm being a stubborn mule :mischief:. I'm investigating it. I have noticed a bug where the AI will still maximize commerce when pursuing a religion - normally that's good, but it shouldn't do so in a wonder building city. That slows down the wonder building. It would also emphasize great people and research in it's best wonder building cities, including while they were building wonders. oops. Now it'll never set those emphasize types while building a wonder. So the wonder stuff, yeah there definitely was a new bug at work there.

Looking at the defense stuff, the AI was indeed happy with just 1 defender until a war started, the AI will now add an extra defender to all cities as empire size reaches the following thresholds: 3, 7, 13, 22 cities...

While I was in that code I also greatly reduced the number of defenders the AI keeps in a freshly captured city (it was obscene), which should help the AI to push on rather than getting it's big stacks tied down in the first city it captures.

In terms of stack composition, the AI is remarkably ignorant of what a siege engine actually is (which is also why it's dumb about escorting them - it really doesn't know). The "City Attack" quality value now takes into account whether the unit receives defensive bonuses. I'm going to look into this function more since it might be overvaluing the bombard factor.

edit: Looking into some stuff I've discovered that First Strike gives a whopping +16% to the combat value of a unit which probably makes the AI overvalue Archer and such. I've halved that to the more appropriate +8%.

Also the bombard rate would seem to MASSIVELY overvalue early siege units.

So I've utterly tweaked that so for City Attackers, units which are NDB are valued less, extra movement is valued less, withdrawal chance is valued less, collateral is valued less, Bombard is valued a lot less, Immunity to FS is now valued (slightly, wasn't before at all), raw strength counts for a lot more. The City Attack Ranking with Knight tech now goes:
Treb > Knight > Mace > Catapult > Sword > Horse Archer > Axeman

Before these changes it was:
Treb > Catapult >>> Knight > Mace > Horse Archer > Sword > Axeman

I might now look into getting the AI to get a minimum number of siege units, like 100% bombard a turn...
 
Thanks for telling me I'm being a stubborn mule :mischief:. I'm investigating it.....
:goodjob:

Thanks for taking the feedback in good spirit, i.e. as intended; to help make your improvements even more awesome. :)

I've just finished my 2nd game with the 12-21 build (hey, its a fun way to escape the holiday TV rubbish), and I noticed the same 'oddities' I experienced in my previous game:
-I could build pretty much whatever wonder I fancied.
-I quickly got to the top two in the score chart and stayed there for the rest of the game.
-On two occasions I was attacked by a stack of siege engines without any other escort.
-The two Civ's with whom I had wars, had some cities close to my border with only two defenders, and were easy targets for my forces.

Now here is the twist,-just as I thought I was cruising to another comfortable score victory, 1940 something, I was beaten by a cultural victory by Ramesses II! :eek: I never expected that to happen. Excellent stuff! ;)
 
Those are great news Blake, I'm really looking forward to the next build with all that you just said ^^ I apologize for looking rude, you're not a 'stubborn mule' obviously ;) Thanks again for all your work and dedication to improve the game in every way.

I agree with the last poster too, the AI cultural win is one of the many great additions of the mod, it definitely adds a new twist to the game.

I finished my monarch Shaka game and won easily in 1920 via space race, though Cyrus and Augustus Caesar were not so far behind technologically. Incas tried a cultural win but probably had the idea too late since they were only halfway when I won.

There were several nice AI wars though, and a few promising sea invasions.

I noticed another oddity though, I had a frigate near an enemy city I was bombarding, there were 6-7 frigates in his city and he decided to do nothing with it... A human would kill the frigate even if he looses 1 of his 6 own frigates, I wouldn't have been able to bombard 10 turns long. I guess the AI can't calculate that.

I don't remember if the oddity was there before (?) but I guess so.
 
Looking at the defense stuff, the AI was indeed happy with just 1 defender until a war started, the AI will now add an extra defender to all cities as empire size reaches the following thresholds: 3, 7, 13, 22 cities...

I was considering suggesting that (# defenders should scale based on size). However, I would also like to suggest that AI defender count should probably be weighted:

- holy cities should get an extra defender
- border cities near hostile neighbors should get extra
- core cities can go lighter on defenders
- top cultural cities should get extra
- cities under cultural influence should get more
- coastal cities should probably get a slight extra
- cities on top of strategic resources should get extra
- cities near strategic resources should get more
- cities that can see large armies of non-allies should get more

That would mean on a 20-40 city civ, core cities would have at least 2. Cultural centers would have 4. Holy cities get 3. Border cities should be 3, up to 6+ on hostile borders. A city on top of a strategic resource might get 4-5 while near a resource would get 3-4. Not exactly sure how the numbers work out... maybe I'll muck in Excel in a bit.

In my 2-defender cities, one is a city-garrison unit, the other is some sort of assault unit to deal with pillagers. Next unit is usually healing w/ CG. After that it alternates between defender and anti-pillager.

Maybe pacifist civs would have more defender units (+25% to the count?). Civs next to powerful civs that they aren't friendly with might want to bump up the defender count as well.

I might now look into getting the AI to get a minimum number of siege units, like 100% bombard a turn...

That would be an excellent idea. Now if only you can convince the AI to bombard and assault a city on the same turn instead of letting defenders heal. :D

...

One thing I noticed with the 12/21 build. Not even sure if the worker AI is under your purview.

Towards late-game, I end up assembling a clump of 3-4 workers in a group. Give them the same # of defenders as a border city (usually 3-4 flat-ground optimized defensive units). Group it all together, then automate it (I don't care what they do at this point, and I don't allow them to remove upgrades). This usually keeps the workers from being stolen and provides a roving reaction force to confuse the AI.

The problem in my current game is one where I have a few new cities far away from my empire. Cities that I took in order to force capitulation of rogue vassals. Both of my automated worker stacks decided to strike out across *enemy* territory to get to those new cities.

Automated workers should probably never enter enemy territory and ideally stay away from enemy territory.

Another thought I had (probably outside the scope) is that any automated workers should halt being animated any time that war is declared. Which would put them back in the "need orders" queue for the player to look at and either re-automate or re-task.

Does the AI even understand the idea of defending workers? Or putting 2-3 workers together in a group (to get projects finished faster and allow for fewer worker defenders)?

...

And I agree that the AI should rally forth to destroy ships within their cultural waters more often. I also never see the AI putting ships on sea resources for defense. My usually strategy for ships is:

- at least 1 ship on every food resource (usually balanced atk/def), then 2 per
- at least 3 ships on oil (one attack, one medic, one scout)
- 3+ ship fleets patrolling the border waters or sailing forth

In later periods, it's usually destroyers first (until I get 3 on a spot), then alternate subs/battleships to get up to the proper count. I usually don't put more then 2 on any food resources. They can band together to attack interlopers. Oil patches are usually defended based on the size of the opposition but no more then 6-8 on a huge map in late-game (1-2 subs, 3-4 destroyers, 2-3 battleships, 1-2 carriers). Patrol fleets are usually a bit larger and move around, assault fleets include transports and are even larger yet.

Pacifist AIs should probably boost their defender counts, especially for resources near cultural / holy cities. Resources should probably be prioritized (based on whether they're being worked?) if there aren't enough units to go around.

(But sea battles are a huge topic. So is the AI defense of strategic resources.)
 
Blake/Iustus and team

I just thought I'd give some positive feedback on the city build governor, I'm pretty much using it as default now (I dislike micromanagement) and I'm very impressed. I particularly like how I can add to the build queue without it disabling, the transition is seamless.

I'm interested to know it you intend to allow the governor to build missionaries (to spread within the players civ only), as they have an impact on both happiness and commerce.

I have noticed the debug mode keeps track of missionaries needed for each religion the player has, which is why I ask (I would also like to see this of course).
 
Running the 12/21 build here with 18 civs on huge/fractal with no issues (one isolated CTD the other day that hasn't shown up again). Running with 2GB RAM and a 256MB video card.

Folks who are having issues should probably indicate hardware specs?
P4 2.8G, X800PRO, 1 GB RAM. I was also playing a huge lakes map (overcrowed) with 18 civs. It crashed towards the begining of the loading process - with about an inch or less of green bar on the loading screen before crashing [geeze, that's descriptive!]).

I'm also experiencing this. It definitely doesn't approve of 18 civs.
Actually, after I made that post (I was in Civ at the time - alt-tabbed out of the custom games screen and then back in - and had it it all set up for 17 civs) it then crashed after I posted it with 17 civs, so I don't know. I fired the game up again and did it for 17 civs and it worked fine afterwards though.

Perhaps the map script or other factors play a role as well? I started a 18player game (Archipelago, Snaky Continents) with the 12/21 build - no crash so far (early ADs).
Hmmm. I played 18 player, Huge Lakes map. Hmmm. Maybe it has to do with the starting code for finding/creating new starting locations and normalising the terrain.

...

Regarding the Terra AI cheating: I don't really play terra, but the human player knows that when they play a terra map, they know that there is another 'uninhabited' world filled with barbs. If the AI went at it with exactly the same expectation, it could make for some interesting games.

...

Regarding my current 12/21 game (huge lakes, 17 civs): I noticed a warmonger go into dagger mode with no plan for war. It sat there for a while until it went to war. I am wondering if somehow the dagger strategy could somehow be tied in with going to war so that when the AI has set itself up for a dagger strategy, it then chooses a target (the most appropriate and most rewarding target for it's situation) rather than wait for the RNG to give the green light. That aspect of it seemed disjointed. Once it went to war, well it was cool to watch. :)

Another regarding my current 12/21 game: The info texts that appear go off the top of the screen so that the info for what wars they have decided to commit too disapear and other important information disapears too. Maybe the info could be reorganised so that the more important information could be displayed down below, like what wars they have commited too, etc. When I alt-hover over my name, my power rating disappears off the top of the screen too because it tells me my odds for dogpile war and what ever else that appears. There is also the issue with when you befriend somone and thus have a lot of modifiers, it sends the AI infomation off the top of the screen :(

I don't mean to be picky, the information is fantastic and gives great insight into how the AI figures things out, but I am finding it frustrating in a game with many civs. I also noticed that you took that info out of the foreign advisor screen. Maybe with there being more room to display the information, it could be useful to put the information back there too.

Great work :thumbsup: :)
 
From the 12/12 version -

I prefer to leave chipotle on all the time for convenience. Normally that doesn't tell me anything I shouldn't know unless I ask. That's why I don't want to see "X traded with Y" messages when I've never met either.

I'd be satisfied if the messages appeared in a different part of the screen, so that I can keep an eye on the upper-left for normal messages and ignore the special messages that appear elsewhere.
 
Back
Top Bottom