They do revolt, especially french and brits.
I think on land the AI still has problems dealing with proxies and forward bases.
1 Like
Hello again. The recent improvement made to the mod have been very nice. As per my last post, this AI is giving me a run for my money. However, there are two problems that have recently appeared with the natives AI and could use some quick fix :
- For some reason, the Aztec AI refuses to mine coin. The post game statistic show a grand total of 0 coin gathered after 20min. This skews the army comp and completely prevents it from ageing up past age 2. If just masses coyotes and send them to their death.
- The Lakota AI has a strong hate for the 3 villagers shipment. Often the deck wonāt have any age 1 card resulting in a deluge of unit shipment in age 2 with a weaker eco behind.
All the comments above are from 1v1 supremacy against extreme AI.
1 Like
@MoosGorgon83309 Thank you for the feedback!
Thatās a likely explanation, Malta may be doing a naked FF. Itās worth noting that games online will go like that sometimes, so itās part of the downside of making the AI play like this. I think the best course of action is to try and improve the AIās performance generally while getting rushed. Iāve also been meaning to prevent the AI from building too many military production buildings in the beginning, but havenāt gotten to it yet.
As for depots - maybe. I donāt know if itāll be worth it since the AI wonāt know to avoid them, but I could probably come up with a way to build them and destroy them fairly easily.
With regards to the fixed gun wagons, Iāve tried to fix it already but donāt know why it isnāt working. For the time being itās just not that high of a priority.
@Moonshadow7475 Iām pretty sure heās referring to Mexicoās revolts specifically, not just the standard Age 4 ones.
For the mexico revolts - Itās something I probably want to look at. The problem with AI revolts in general is that they arenāt very skilled at maintaining momentum, and for whatever reason players often like the long drawn out age 5 gameplay.
@ReegarGW2 Thanks for noting the issues! Iāll see if I can replicate them and figure out whatās going on. Lakota should always have 3 vils, so something is wrong if they donāt.
About forward bases, we figured out that in treaty after use spy we can see that they try to do it all at the same spot so they are not very effective at it. (we saw patterns overlaping)
We will have to try it out more, as we try between different AI mod and classic to compare ^^
Are they al trying to build as soon as the treaty ends, at the same time?
1 Like
Update. Just to be sure this was not a āmeā issue, I removed the Hammerhead AI files left on my drive and ran a file integrity check. The Lakota AI now uses the 3 vil shipments. It was a āmeā issue. Sorry for the false alarm. ^^ā
1 Like
Oh nice! I wonder if youāve been playing against hammerhead this entire time?
Nah 'cause when I disabled assertive AI prior to the file deletion the age-up times changed for the AI. I might have played an unholy hybrid.
I tried some otto FI against lakota tonight and the AI sent the 3 vils (that was new), then a deluge of cavalry in my base (as usual).
New update is out 2024-08-30T07:00:00Z!
This is a pretty cool one, the AI now has the ability to raid villagers on non-island maps separately from itās standard attack plans. Different personalities will raid more than others, so give Gall or Frederick a shot if you want to see the new feature (90% chance for those two).
Thereās also a couple more fixes, like the AI is more likely to rush in 3v3ās and 4v4ās.
Hereās an example of the raiding:
I lost about 10 vils to raiding, but of course itās not just the villagers lost, itās all the idle time and not being able to freely gather from natural resources. I hope yāall like it, itās been way harder for me playing vs the AI with this change, at least 1v1.
4 Likes
Thanks for continuating the update !
Some notes for treaty games :
1- As native AI, Napoleon is doing the French Revolution and donāt know how to use it (no Napoleonic Era, no Sans Culotte Eco)
2- As native AI, with too much money they are too prone to mass artillery (12-20ish in all context, sometimes all mortar in defensive position hum) limitating the use of other counters.
3- AI deck in age 4 have still a lot of one use units cards, especially non-europeans. Age 2 decks value too much gold mines and hunts improvements (Spices trade and silversmith)
4- Didnāt notice before that they had built-in bonuses (like 35%), is that new ?
5- They are doing good shapes walls in some place of the map (very inconsistent behavior, generally 1 a game vs 4 AI), but if they try at first to put their building in it they let it go easilyā¦so itās weird ^^
6- Pretty stable overall, not lag during even long game so that is a good point 
We will try some more games out with this mod. Good job ! 
1 Like
Thanks for the notes!
- I couldāve sworn I looked at this but it wasnāt a straightforward issue. I would rather get the French revolution working, though thatās a lot more work than just removing it.
- Iāll look into it at some point, but in the meantime I recommend you stop massing towers and infantry.
- Sounds like a treaty deck thing. Iāve been meaning to look into this again, but it isnāt a huge priority as long as there isnāt something glaring, like Aztecs picking less than 25 cards.
- Itās 15% and 30% for the two hardest difficulties, same as the standard AI. Handicaps can become a touchy subject so I would much rather leave it as-is. What I did do, is make it so the top 3 difficulties all play the same, except for the handicap. Hard with a 30% handicap will play exactly the same as extreme. If you find yourself needing more of a challenge you can always give them a handicap in the skirmish setup.
- Iām not sure what you mean by inconsistent. Thereās a bit of variability built-in so not everyone should be making forward bases in the same way at the same time (or at all), but thereās also some technical limitations with how itās done and I wouldnāt be surprised at all if weird stuff is happening on tight maps and odd team starts.
- Excellent!
About forward bases, they tend to be like just in front of our bases, without any in-between bases, so in large maps itās can seems a bit weird.
As well I was looking to a 3v3 AI games, and I see AI doing a lot of large going in and going back to his bases with a train of units.
(Like āKill a forward tower, ok I go backā without pushing forward.)
In large maps they roaming in the map more than doing big push (they do, but not that often at the end ?)
Itās may be a design choice so they donāt stomp at us to hard ?
They may are not veryā¦optimized for large maps ? Just asking.
1 Like
Iāve had an idea to try and creep forward bases forward in the ultra-late game, but itās not a very high priority. I would much rather focus on some other things, like the second thing you mentioned:
I think thereās a couple things going on. I think the train is a bug I may have inadvertently added when I made bases bigger. The trains of units arenāt going on an attack, they are part of a defense plan, likely ādefendingā all the way into your base.
The big attacks on trade posts are another issue. AI loves to attack trading posts and I think itās because it calculates it like a free win, but I would need to dig into it more. Iāve figured out simultaneous attacks with the raid plan, so it is possible to send 5 pike to go kill a TP and send the main force somewhere else. Plus, Iāve found that getting multiple things going on at once is one of those things that really raises the intensity of playing vs the AI.
Updated 2024-09-20T07:00:00Z, nothing too major just some incremental changes/fixes, mostly to Archipelago behavior.
2024-09-21T07:00:00Z More incremental changes.
2 Likes
Hey,
Played and watched a fair few AI games.
-
Still not great at judging strength of treasure guardians. Dies a fair amount of the time to them like the base AI. Iām not sure how it decides which guardians to attack but I think guardians do come in ātierā levels. Either way, itād probably be better off only going after really easy guardians or not doing so if it has less then 50% health. Otherwise it just ends up helping the player or another AI by āweakeningā a hard treasure.
-
Probably the biggest weakness of the AI in general is the insistence on attacking against overwhelming odds. It never retreats from an engagement it has no chance of winning. More of an issue in 1v1. Team games though have the whole issue of not helping each other against heavy aggression. So while rushes do āworkā they probably shouldnāt.
-
I appreciate the attempt for AI to use livestock but Sheep are not worth their investment. Unless they get buffed, I wouldnāt bother with them unless they send ranching and use cows which are worth it and might actually be a good idea based on the Ais strengths. Also, they donāt seem to research the fatten upgrade. Not every AI does this though.
-
I wonder about the flavours and whether this is a good idea. The AI doesnāt do early aggression well unless itās a team game which as said previously has other issues.
-
The āstoneworksā the AI sometimes attempts to place in the middle of the map usually just lures other AI into attacking the walls pointlessly. It doesnāt always seem to do this though.
-
It probably should move onto mills at least building one or two earlier. Maybe this is what the livestock thing is for but again not every AI personality does it. Sheep are still an awful investments though and Iāve only seen a few AI do it. In any case its very common that it has no food income at all if it is under any sort of pressure and wastes loads of villager seconds walking back and forth between the hunt, getting attacked and going back, killing the units attacking them, going back to the hunt and then going back again.
-
The AI seems to consistently delay building a market even more so then the base version. The market upgrades are very much worth it, especially the first ones in age 1 and steel traps in age 2. The others take a fair bit longer to pay for themselves. Sometimes the AI doesnāt build markets as late as 10-14 minutes into the game. If you are investing in other forms of income like fishing boats, trade posts, banks, manor houses, markets can wait but usually they arenāt. For the AI who canāt currently spend resources in super efficient build order style where every resource is tightly budgeted, market and hunting dogs should be a must very early. Even as early as the start of the game for nations that start with enough resources to do it without many problems.
-
It will sometimes send unit shipments and do nothing with them. Unit shipments have two purposes, attack with a timing or defense. If it isnāt planning on attacking, it probably should send crates or economic cards and build the units manually instead. It could have town militia in its deck (if applicable) if it does come under attack if itās possible to make it do this.
-
The AI often advances to the next age attacks while it is aging or just after and doesnāt take advantage of the age by getting upgrades or anything and often just throws a smaller army away against someone who hasnāt aged up. This might be hard to deal with, I only have a limited understanding of how the AI works.
-
Last thing for now. May not be your focus but the AI doesnāt take advantage of certain civs. Some of the decks while better then the base, could still use some work.
2 Likes
Thanks for the feedback!
- This is controlled by the exploration plan, meaning the only way for us to fix it is to write it all from scratch.
- The AI really sucks at retreating, especially against players. The consensus had been to just let the AI fight to the death instead of retreating too early. This is something I want to address if I get around to rewriting attacking in general.
- I can take a look and see if I should reduce the amounts.
- Iām pretty committed to this. I know some playstyles will do better than others, but the solution is to improve those playstyles not do away with the whole concept.
- Thereās an issue I know about where the AI looks like itās attacking, but itās actually defending and the defense radius is just way too big.
- I can take a look, but the solution is probably to defend the hunts with a tower or something. The old issue was the AI would switch to mills way too early when they had hunts right outside their base.
- Iāll take a look when I look at Italy. Might as well just write the market rule for everyone.
- This is some tricky stuff to get right. Itās probably shipping units because it wants to attack but isnāt calculating a worthwhile attack so thereās no obvious solution.
- Team or 1v1? In team games I have them just wait until most of the players have reached the new age. I did that since one teammate would lag behind, and then by the time they made it up the other teammate would start advancing so you would wind up with the AI effectively doing nothing until Age 5.
- I would need specific examples. I did have an old idea to try and implement build orders but I never got the system fully working.
1 Like
Itās been a while, but thereās a few changes:
- naked FF improved
- over island base building improved
- fixed an issue preventing ships from ungarrisoning from docks
1 Like
This update causes the ai to fail to load for me.
1 Like
Thanks, it should be fixed now.
1 Like
working fine again, thanks!
1 Like
2024-12-06T08:00:00Z
A couple important fixes:
- Each AI will only attempt to build 1 star fort
- In cases where the ending age was lower than a strategy, the AI will no longer use that strategy.
- i.e. if the ending Age is the Commerce Age, then the AI will not use the Fast Fortress strategy
- (Useful if youāre in the mood to recreate ESOCās current tournament rules)
4 Likes