[BTS] v0.11 Discussion thread

[jdog5000]

Post any and all questions, observations, critiques, etc on Better BTS AI v0.11 here.

Better BTS AI Changelog for v0.11
_________

- Fixed bug in player closeness calculations causing the AI to be blind to its actual proximity to other players (CvPlayerAI::AI_playerCloseness, used Option A from Unofficial Patch discussion thread)
- Fixed bug with needed defenders where the AI would choose to defend a recently captured holy city less than a normal city
- Modified how closeness is used for determining who to attack, different settings for regular and Aggressive AI
- Modified city closeness calculations to use population from both cities, not just pop of the other city
- Created functions for determining when cities share the same water area (lake, ocean)
- Lowered demand for transports when either in a land war or when the civ can only build galleys
- Partial implementation of a new A* solver for determining closeness, does not yet affect calculations
- Modified debug mode output to show relevant information for these developments
- Added AIAutoPlay mod
- Removed unnecessary variable saving from AIAutoPlay to preserve save game compatibility with vanilla BTS (NOTE: save compatibility now broken with v 0.1 but will be maintained in future versions ... sorry)

Better BTS AI also includes Solver's unofficial 3.17 patch with the following changes:

* Fixed unavailability of foreign civ demographics if playing with espionage disabled.
* Barrage promotions made working again on Tanks and other units with no base collateral ability

 

[glider1]

I think in build 0.11 of Better BTS there should be a closer look at the size of attack stacks the AI sends towards a target immediately after a border war has been declared. My test conditions are:

Terra, temperate, monarch, 18 civs, 120 turns in, standard or aggressive.

I think it's good on standard aggression but the probability of attack stacks capturing a border city on aggressive seem a little too low I perceive. However the degree to which this observation is true or false is only quite small (10% say).

Possible reasons for this is a slightly "premature" tendancy to declare war on aggressive before the attack stack assembly is of sufficient strength. Not sure! Keep an eye on it....

Cheers.

 

[MaxCiv]

Would it be possible to have a "Sticky thread" for new versions? It would make it much easier for people to find the latest build. Thanks for all the hard work.

Max

P.S.
Solver just updated the unofficial patch. Are you planning to keep Better AI in sync with Solver updates? I wouldn't want to be forced to chose between the two

 

[jdog5000]

I think in build 0.11 of Better BTS there should be a closer look at the size of attack stacks the AI sends towards a target immediately after a border war has been declared. My test conditions are:

Terra, temperate, monarch, 18 civs, 120 turns in, standard or aggressive.

I think it's good on standard aggression but the probability of attack stacks capturing a border city on aggressive seem a little too low I perceive. However the degree to which this observation is true or false is only quite small (10% say).

Possible reasons for this is a slightly "premature" tendancy to declare war on aggressive before the attack stack assembly is of sufficient strength. Not sure! Keep an eye on it....

Cheers.

I'll check that out, I've been testing standard so far and been very happy with war decisions and execution. The one question mark I've seen is that when the AI does decide to join an intercontinental war, they may send a huge first stack but they don't seem to follow up by shipping over more troops. I think this may have to do with AreaAIs ... I'll post about it in the strategy thread once I do some more investigation.

Would it be possible to have a "Sticky thread" for new versions? It would make it much easier for people to find the latest build. Thanks for all the hard work.

Max

P.S.
Solver just updated the unofficial patch. Are you planning to keep Better AI in sync with Solver updates? I wouldn't want to be forced to chose between the two

Yeah, at some point we'll get organized and set the forum up properly and get a moderator to sticky/unsticky things.

As for the unofficial patch, yes I intend to keep things in sync ... many of the SDK updates for 0.15 came from this mod I think the barrage for armor change is XML based so you can just take the XML sub-folder out of the unofficial patch for the time being.

 

[jkp1187]

jdog--

Are asserts enabled?

 

[glider1]

I'll check that out, I've been testing standard so far and been very happy with war decisions and execution. The one question mark I've seen is that when the AI does decide to join an intercontinental war, they may send a huge first stack but they don't seem to follow up by shipping over more troops. I think this may have to do with AreaAIs ...

I concur, so far in the middle ages during land wars on standard aggression, I cannot immediately fault AI war decisions and execution now, at least from a "basic" perspective. The AI will DOW on a non-neighbour in the presense of "neutral" neighbours. The invasion seems to be near enough (say <15 tiles) to warrant such an offensive declaration. The AI will bring in sufficient troops to capture an outpost city and then begin pillaging from there. I have one example I am examining under the microscope right now.....

These seems to be signs that an AI might even be able to win a domination victory one day if it can only learn how to properly conduct an intercontinental invasion! From memory, I have never experienced an intercontinental invasion that wasn't simple a single tile big stack of navy and invasion troops arriving in one location....

It would be very interesting to understand the mechanics better....
Cheers.

 

[jdog5000]

jdog--

Are asserts enabled?

No, it's a Final Release build so no asserts. I could make one as a separate download if you want.

These seems to be signs that an AI might even be able to win a domination victory one day if it can only learn how to properly conduct an intercontinental invasion! From memory, I have never experienced an intercontinental invasion that wasn't simple a single tile big stack of navy and invasion troops arriving in one location....

It would be very interesting to understand the mechanics better....
Cheers.

Yeah, I'd really like for invasions to work well ... as you said, I've seen large initial stacks but never any substantial follow-up of ground troops. Once the transports get back to the main area they wait and aren't refilled or at least not nearly fast enough.

 

[jkp1187]

No, it's a Final Release build so no asserts. I could make one as a separate download if you want.

I think that would be a good idea. It's what Blake and Iustus were doing when they were working on the original Better AI back in '06-'07 and I think it helped them identify and quash a few bugs that might have otherwise gone unnoticed.

 

[glider1]

Just a thought about the way to control "improvements" in each new build of "Better than BTS". There could be option control system in the build for each "improvement" so that various improvements can be turned off or on. This would help testers focus on particular "improvements" that have been made and would also help control and limit the interactions between "improvements". However that is only possible where an "improvement" is highly localised in the code and not smattered throughout a thousand places like is sometimes unavoidable.

The XML system allows for on the fly updates of data and so options can be turned off and on via GlobalAltDefines.xml without having to reload the game or the save. I would have been reluctant to go down that path before because I noticed that on the fly XML reading was occasionally prone to a CTD. But in 3.17, Firaxis have apparently made some "improvements" on XML reading for modders.

Something to consider.
Cheers.