2.5 Preview
- Thread starter Tholal
- Start date
Very nice version. Many good things and better AI performance in general. They can really win religion and mastery for sure now. A few comments if you care.
Played two games. In one, after building an advantage with the bannor, I let the game go on AI auto play for 100 turns.
Then, the bannor ai placed, until the end of auto play, a stack of 26 summoning mages on a remote city and flagged it as perma defense, which should probably not happen, away from the area of skirmishes.
After making peace, the bannor ai went for culture vic strats as a priority and spammed all kinds of missionaries for some reason, making every city have 3 to 5 religions. If the spam is related to culture strats or an overvalue of temples or whatever, i don't know, but it can't be right, either for culture or the temples in the late game.
Another game. Built advantage with Doviello, then auto play, with 3 vassals against a big Elohim with 2 vassals including Mercurians (The Elohim went hardcore OO, then harcore AV religion, which was funny; still they did not summoned Hyborem: they can't do it after the mercurian gate, I can't remember?).
First, even with +18 friendly I had a hard time trading with the Elohim. I understand that you don't want humans to exploit the AI, but, although it's ok to be stubborn trading mana and strategic resources, in relation to the rest of the other resources they should be more flexible, at least when friendly.
The peace defenders Elohim declared then at +18 friendly against the Doviello, although they were in two other wars. Now here's the thing good and bad: The Elohim had the 4 towers and were ready to build the tower of mastery. But they didn't do it. They played a defensive war losing cities slowly and switched to culture vic strats(?). The Doviello AI then went on to impress me building very quickly all 4 towers using dispel left and right and rushed the tower of mastery, all while pushing back the elohim and angels (they still sucked at taking cities with many defenders though despite the numbers advantage). So, for the first time I saw not just one but two AIs have the 4 towers. The Elohim fell short at the end with silly strat switching. It was weird.
Anyway, I think one should play with culture option off, right now.
Also the Doviello vassals broke open borders about 3 times - should never happen, makes no sense.
Stacks of mages with metamagic I unfortunately waste the casting turn spamming floating eyes - from the same tile. I think I was seeing like 12 floating eyes at some point every turn.
Finally, I left the Doviello in republic, guilds but they switched to arist, conquest. In a boreal map (true, a bad map, but playable) with mostly tundra and plains tiles, that lead to 2 food farms . The AI does better now abusing aristo in general, but in the late game they should value guilds and specialists a bit more with the bigger happy cap, and not just the phi leaders.
If it helps in some way.
Played two games. In one, after building an advantage with the bannor, I let the game go on AI auto play for 100 turns.
Then, the bannor ai placed, until the end of auto play, a stack of 26 summoning mages on a remote city and flagged it as perma defense, which should probably not happen, away from the area of skirmishes.
After making peace, the bannor ai went for culture vic strats as a priority and spammed all kinds of missionaries for some reason, making every city have 3 to 5 religions. If the spam is related to culture strats or an overvalue of temples or whatever, i don't know, but it can't be right, either for culture or the temples in the late game.
Another game. Built advantage with Doviello, then auto play, with 3 vassals against a big Elohim with 2 vassals including Mercurians (The Elohim went hardcore OO, then harcore AV religion, which was funny; still they did not summoned Hyborem: they can't do it after the mercurian gate, I can't remember?).
First, even with +18 friendly I had a hard time trading with the Elohim. I understand that you don't want humans to exploit the AI, but, although it's ok to be stubborn trading mana and strategic resources, in relation to the rest of the other resources they should be more flexible, at least when friendly.
The peace defenders Elohim declared then at +18 friendly against the Doviello, although they were in two other wars. Now here's the thing good and bad: The Elohim had the 4 towers and were ready to build the tower of mastery. But they didn't do it. They played a defensive war losing cities slowly and switched to culture vic strats(?). The Doviello AI then went on to impress me building very quickly all 4 towers using dispel left and right and rushed the tower of mastery, all while pushing back the elohim and angels (they still sucked at taking cities with many defenders though despite the numbers advantage). So, for the first time I saw not just one but two AIs have the 4 towers. The Elohim fell short at the end with silly strat switching. It was weird.
Anyway, I think one should play with culture option off, right now.
Also the Doviello vassals broke open borders about 3 times - should never happen, makes no sense.
Stacks of mages with metamagic I unfortunately waste the casting turn spamming floating eyes - from the same tile. I think I was seeing like 12 floating eyes at some point every turn.
Finally, I left the Doviello in republic, guilds but they switched to arist, conquest. In a boreal map (true, a bad map, but playable) with mostly tundra and plains tiles, that lead to 2 food farms . The AI does better now abusing aristo in general, but in the late game they should value guilds and specialists a bit more with the bigger happy cap, and not just the phi leaders.
If it helps in some way.
Does the AI know how to bribe others into changing religion when going towards that victory?
Also, Samhain spawned units should probably be immobile for the first turn they appear. You can't do anything to prevent frostling wolf riders from capturing workers or pillaging stuff since they spawn and act between your turn. Since they can appear in fog busted tiles, that can't be right.
Also, Samhain spawned units should probably be immobile for the first turn they appear. You can't do anything to prevent frostling wolf riders from capturing workers or pillaging stuff since they spawn and act between your turn. Since they can appear in fog busted tiles, that can't be right.
Definitely appreciate the feedback. Do you happen to have savegames of any of the specific incidents you mention? In particular, the stack of 26 mages as defense and the various Elohim issues.
Looks like it's related to the BUG unit naming option? Did you have that option enabled?
Not that I'm aware of. Diplomacy is something that I really haven't done anything with. I'll add this to the request list.
Unfortunately, I wasn't planning on posting so I don't have the save with 26 mages on permadefense.
I have a save somewhat before the doviello auto play; but it doesn't mean that if I post it (I can post it if you want, though) and you download it and auto play it that you'll get the same results as I did. The second time I auto played it from the save the Doviello smashed the Elohim capital on the north that had one tower and eventually won domination in about 90 turns. That did not happen at all in the first time: the doviello just took some cities on the south and, although the elohim were ahead with 4 towers and the doviello none, the doviello were the ones that won a mastery victory like I said.
Btw, the Elohim with evil alig wasted time researching righteousness. Techs like that and malevolent designs should only be researched with the right alig and conditions, otherwise a huge waste of beakers.
You are correct, I did not notice it. I think this issue had come up before in the forums.
I asked that because, after the improvement to ai inquisitors, I guess one of the few things left for the ai to accomplish that victory more effectually is to know how to bribe others.
In all of the other victories paths the AIs are doing good (with the exception of culture, that they never get or get close at) as long as they don't divert and speculate in other directions once they reach a certain stage in the specific victory path.
Other things:
If I worldbuild a few hawks in a city and hunters and auto play it, the ai does not grab the hawks with hunters (probably known).
Firebows should have more unitai than the longbowmen. I just autoplayed a stack of firebows with bambur and a few sok and instead of helping to capture the last enemy city they just dispersed back to garrison cities leaving bambur and company to just pillage.
The ressurection spell is virtually useless since the ai knows how to explore graveyards, so no graveyard survive to the late game. I vote that graveyards inside someone borders cannot be explored by others, so that at least there's a chance of that spell being used by a human.
Freesmog
Skeptic1938222569
I agree. Seems like the AI isn't always picking the right techs at the right time.
Regarding graveyards, that's a good point but I don't know how much the AI would actually use them. I do agree that other civs should not be able to explore anything in another's territory. That's like allowing them to pick up items, but in this case it also means anywhere on the map
Here's the mage thing again in another game. In the first save it went into auto play and the mages were soon put into permadefense. In the second save, 110 turns later, the mages are still doing nothing in the south although the bannor ai is getting crushed in the north.
I'll look into it.
Yep. Though I think it's actually the other way around. The hawks would have to load or rebase onto the hunters. But first would be getting the AI to build Hawks!
Good point. Will add more UNITAI options for Firebows in 2.51.
Resurrection doesn't require a graveyard.
The issue here is that you're starting autoplay in the middle of a game. Whenever an AI player creates a unit, that unit is assigned a UNITAI based on what the AI player wants to do with it. When a human creates a unit, the unit is simply given the default UNITAI. In this case, the default UNITAI for Adepts is UNITAI_MAGE, which the AI players use for city defense. So when the AI takes over, it sees that these mage units have UNITAI_MAGE, and sends them off to guard a city. It would be nice if a unit could evaluate its current AI type and decide if it needs to switch to something else, but aside from a few special cases, there's really nothing like that in the code. I would like to add a function for that at some point.
Multiple adepts guarding a city is another issue in of itself. On my bug list.
I think that the Doviello AI should be finally prevented from building workers as means to upgrade to an army, especially now that beastmen don't get obsolete. It's a waste of hammers/food since, of course, workers are more expensive than beastmen.
AI is not building enough champions in the late game. They do build tier 4, but tier 3 melee units sometimes are not really seen.
Imho, AI should be prevented from building warriors in the late game; indeed they build too many throughout the game: it does come a point in the game where warriors mean nothing but free exp.
AI is not building enough champions in the late game. They do build tier 4, but tier 3 melee units sometimes are not really seen.
Imho, AI should be prevented from building warriors in the late game; indeed they build too many throughout the game: it does come a point in the game where warriors mean nothing but free exp.
Agreed. And AI upgrades warriors to longbowman.... sitting ducks
question : do worker upgrade to beastman or to any other melee unit ?
in the second case, du to the hammer worth, it will cost less to upgrade to axe than from beastman.
and still workers have 4 advantages:
-they do not cost military upkeep nor military supply out of territory
-they can be built faster due to food production (even without the conquest civic)
-they can render themselves useful during wait
-they move 2 tiles at a time (very useful to go to the front, escorted by 2 hunter or 2 pack wolves)
so, maybe the AI should be bared from it.. but IMO, it is only until someone finds the time to teach it how to profit from it.
IN any cases, maybe workers should be able to upgrade to any melee unit as long as tech is known, without stopping by the step "beastman" and it might help leverage them. (and to any unit beastman upgrade to)
in the second case, du to the hammer worth, it will cost less to upgrade to axe than from beastman.
and still workers have 4 advantages:
-they do not cost military upkeep nor military supply out of territory
-they can be built faster due to food production (even without the conquest civic)
-they can render themselves useful during wait
-they move 2 tiles at a time (very useful to go to the front, escorted by 2 hunter or 2 pack wolves)
so, maybe the AI should be bared from it.. but IMO, it is only until someone finds the time to teach it how to profit from it.
IN any cases, maybe workers should be able to upgrade to any melee unit as long as tech is known, without stopping by the step "beastman" and it might help leverage them. (and to any unit beastman upgrade to)
Broken Hawk
Emperor
Hah!!! I'm late to the party. I've been waiting the release of 2.5 on the original thread. Can't wait to try it out. As always, thanks Tholal.
jackal1234
Warlord
please more update release.. i know its hard to do with two little ones, but i was honestly surprised and found c5 brave new world semi-playable despite my deep horror on legit purchase of initial c5 crapola alpha retail when it first came out... thanks for the good time!
