Version 0.84 discussion thread

[Andy06r]

Sorry for double post but the AI is effectively broken and is unplayable on archipelago maps using this mod.

8 AI's, standard arid archipelago. I already showed you in my previous post some questionable city suggestions (as far as I understand, the AI uses those suggestions itself)

Here are some more screenshots ...

It is turn 76 ...

2 AI's have not built a second city
The other 5 AI's all have second cities of size 1 or 2

As you can clearly see, Boudica has a great island but refused to expand - yielding her copper to me.

Isabella apparently decided that now is an amazing time to build the oracle, despite only having 1 city

And Egypt thinks that with 2 cities, size 4 and 1 that now is a great time to build the great wall - except this is an archipelago map

It almost seems like all the AI's tried building the same wonder and none of them switched off it - or something *shrugs*

 

[Fimbulvetr]

Could you upload a worldbuildersave of turn 51 for me? I know the AI founds cities in places a human would never touch but that one is indeed a bit much.
And maybe a save for first turn too.

@PoM: did you try that improvement happy code yet?

edit:

The only way this could happen is if the city's computation of what the best build for a plot is changes while the first set of workers are working. Or, I suppose, if which city is working the plot changes. Should be possible to check when setting city best builds for a plot whether something else is being built and cancel it.

Actually I think it is not supposed to happen at all: take a look at the code dealing with eForcedBuild - maybe you can figure out why that doesn't stop bestplotbuild to change when a worker is already improving the tile.

 

[Andy06r]

Here is the first turn savegame - I no longer have anything later, sorry - there isn't a new seed on reload though.

 

[Andy06r]

I apologize for the double post, didn't realize I couldn't edit.

I started another archipelago game and this is the turn 51 save -

I was japan

*No AI has built a second city
*Only one is attempting to
*One of them is building a barracks (this early in archipelago?)
*Two are making lighthouses even though both cities are angry
*3 are spamming warriors/archers
*1 is building a galley, which is arguably smart because of their start

But lighthouses and barracks before a second city? Do you really need 4-5 units on an archipelago before your second city?

 

[Fimbulvetr]

I don't care for any other turn 51, I wanted the one to go with the screenshot. But since you not only don't have that anymore but also seem not to know what a worldbuildersave is, nevermind.

 

[Andy06r]

I wanted to make sure I had the correct save so I loaded the turn 1 autosave, not realizing that would knock out all future ones ...

This "problem" is easily repeatable, make any medium sea level arid archipelago with 8 civs, raging barbs and without fail over half of the AI's will go brain dead. I refuse to play that map script anymore with this mod because it simply doesn't work ...

I'm sorry for not knowing the first thing about modding, but this is the feedback thread and my favorite map script is unplayable - that's feedback

 

[Fimbulvetr]

Just to clear up any misunderstandings:
I have no problem with your feedback (it's not like I'm the one maintaining Better AI anyway), I have observed AIs neglecting initial expansion in favor of wonders or even missionaries often enough, not sure why though, opportunistic wonder and missionary production logic seems to be unchanged, the way they were in BTS.
But since I already know about it I was not really interested in that aspect, even though it seems to be even worse on archipelago.
I was interested in that recommendation of the dessert tile with no food around it as best plot for your next city because that really makes no sense to at all.
But since I'm not using standard Better BTS AI your savegames are useless to me, you'd need to enterworldbuilder and save the scenario from there.

 

[jdog5000]

Andy06r -

Sorry to have busted the early expansion on archepelago ... I'll try that map setting and see if I can figure out what change might be having unintended effects.

 

[jdog5000]

Figured it out ...

For the AIs first settler build I added a clause intended to cause the AI to build defense to escort its first settler before the settler itself. It wasn't written well and instead cause the AI to only build a settler if it already had two defenders in the city. This puts a big delay into the AIs preparations ... this effects all maps unfortunately, so definitely thanks for speaking (typing?) up and getting this addressed!

Checking out a few other issues with archipelago starts now as well, the AI seems to go a little heavy with the workboats early because of the prevalence of water resources around its start.

 

[Andy06r]

Thanks! I've followed this mod for quite some time now (I started using blake's AI with regular civ and then switched to this when bts came out) but I don't post much.

Sorry about getting defensive, that wasn't constructive.

 

[Fimbulvetr]

If I understood it correctly the idea was to build the escort before the Settler, and I don't see how this was a bad idea - unless you can somehow determine when it is safe and make the settler move to its hopefully not completely ******** "best found" destination on his own.
Also it is (or was, if r537 is current) definitely possible for the AI to build a settler as second build, first build is always unitai attack (why not city defense?) warrior, and the turn the production finishes there really is a warrior with unitai attack standing in the city. Now the settler 1 code only checks if unitai city defense units == 1, and since the number is 0 at that time, a settler can be built. (The warrior changes its unitai to city defense the same turn but after the build is chosen). Of course once built the settler will lack escort so it will have to wait again, which is what causes a lot of delay. In the latest rev it's just bLandwar controlled, so can a settler now move without escort?

worker first build:
With that somehow forced attack/defence warrior first build (for noble at least) there is no chance for a worker first build, even if there is a bonus that could be improved with starting techs. If it's not a starting tech it wouldn't detect any bonus as improvable for worker 1 even if the needed tech can be researched and takes less time than the worker, since at turn 0 no tech is yet chosen; for worker 1a I'm simply not sure if the bestbuilds are even set when the first production is chosen, and there's the issue with the city radius still being at 1 at the time, even though it will always be 2 by the time the worker would finish.

I see no problems with going for workboats for seafood in BFC though, unless that means neglecting land workers.


Improvement oscillation: cottage-town and workshop
Maybe simply valuing the invested turns higher does the trick, at least to protect villages and towns. I'll try replacing
iValue -= kImprovementDowngrade.getUpgradeTime() * 8;
with
iValue -= kImprovementDowngrade.getUpgradeTime() * ((pPlot->isBeingWorked())? 20 : 10);
and hope this won't break anything.

 

[LunarMongoose]

On a related note, may as well report something a little flapped up that happened at the beginning of our current MongooseMod test game, which uses BBAI 1.84b:

Ragnar had spawned relatively close to me. Because of the Stone Age stuff in my mod, initial expansion is a much slower process than in vanilla, but in any case he declared war on me quite early, at a point when I had something like 3 cities built and he had 2.

From there onward it seemed like he was no longer willing to expand even though he had lots of room to do so, basically. He threw a few stacks of units at me, but given how early it was they weren't all that difficult to defend against, so I just let him keep suiciding on me until quite a bit later when I was ready to steamroll him with much more advanced units. His tech progress had almost halted because he never built any more cities, and still had 2 when I had gotten up to 8 cities or so (doing good research, ie I hadn't overexpanded too rapidly).

So anyway, just tossing this in your pile.

 

[clif9710]

I've just built up the courage to install various mods. Currently I've got .84 loading with the game (I hope!) and then I select BUG as a mod.

I've noticed that when BUG loads it is evident on the splash screen, where I see the word BUG, then python loading etc. I don't see any reference to .84 on the splash when the game first loads, however.

In game, how can I confirm that .84 is active?

 

[NotSoGood]

I've just built up the courage to install various mods. Currently I've got .84 loading with the game (I hope!) and then I select BUG as a mod.

I've noticed that when BUG loads it is evident on the splash screen, where I see the word BUG, then python loading etc. I don't see any reference to .84 on the splash when the game first loads, however.

In game, how can I confirm that .84 is active?

When running the game, hover over your flag and there should display what version of Better AI you're running.

 

[clif9710]

Yep, it's there. Thanks!

 

[Cybah]

is it possible to check if own cities are under high culture pressure? the AI should conquer or destroy high culture cities near own borders - or the vassal's cities will keep the culture pressure and reduce the own culture during "peace".

 

[Newcomer24]

In my last game an AI founded both Buddhism and Islam... as soon as it had a great prophet decided to build the holy building... a smart decision would have been to pick the religion with the higher diffusion... Buddhism had about 50% world diffusion while Islam only 5%... no brainer I would have said... needless to say the AI picked Islam

Is it possible to add a check for when then AI has a great prophet and decides to build the holy building so that if it founded two religions the one with higher world diffusion is picked?

 

[Yakk]

You mean world coverage (ie, 5% of cities have Islam, 50% have Buddism?)

 

[Newcomer24]

You mean world coverage (ie, 5% of cities have Islam, 50% have Buddism?)

Yep sorry my english sucks at times.
I mean why you should build the holy building of a certain religion if another religion you founded has a lot higher world coverage and it's also your state religion?

 

[Cybah]

it's also affected by the favorite religion.