Page 1 of 1

Does the AI properly consider outpost colonization?

Posted: Fri Mar 24, 2017 7:14 pm
by mem359
I've attached an image of a nearby AI that made some peculiar choices.
It put outposts on potentially "Good" planets, but then didn't bother to colonize them.
That seems odd, considering that they were 1 jump away, and it is now turn 100.
(I didn't start attacking until turn 80 or so. There was a nearby Sentinel, so maybe that caused the AI to go exclusively military?)

The AI is Chato, and Kephra is the homeworld.
Stern Alpha has two Toxic planets, which would be "Good" for the Chato.
The AI did conquer Stern Beta, which had a Mu Ursh colony. It also built two colony bases there (so 3 planets with Mu Ursh).
I did not find any Chato colonies (other than the homeworld).

Re: Does the AI properly consider outpost colonization?

Posted: Fri Mar 24, 2017 8:47 pm
by Morlic
While there seems to be some issue here, it is impossible to say what is going wrong based on a single screenshot.

Can you provide any of these?
- AI log
- Savegame of the screenshotted situation (or some turns before...)
- Savegame of first turn

If you have all of those, that would be great but any would help.

Re: Does the AI properly consider outpost colonization?

Posted: Sat Mar 25, 2017 2:32 am
by MatGB
Not noticed this significantly for awhile, but I am noticing in my current game that I'm going up against a LOT more uncolonised outposts than normal, the AI I've just finished demolishing had a lot more outposts than it had colonies, I think there might actually be some sort of very recent glitch, I'll get some logs and a savegame organised: I am playing with a few test commits from others but they shouldn't make a difference overall.

ETA: 2 autosaves, oldest and most recent, plus the relevant AI log. It's a Trith empire and I'm playing with the Species Mix branch changes which you'd need to load it up else they'd die instantly (Radiated homeworld is a change).

Re: Does the AI properly consider outpost colonization?

Posted: Sat Mar 25, 2017 8:36 am
by Morlic
Thanks, Mat. That was very helpful information.

AI handling of resurrected species got broken 19 days ago. It assumed it would be able to colonize with Misiorla and wouldn't waste time colonizing with some useless Trith.

That particular issue should now be fixed.

Re: Does the AI properly consider outpost colonization?

Posted: Sat Mar 25, 2017 8:14 pm
by Dilvish
thanks for catching that, Morlic, last I had seen of that extinct species PR there was still discussion about just what change was going to be made, I didn't notice when the final change was made and it got put into master.

Reminder for everyone: when you change things that the AI uses, if you aren't up to adapting the AI yourself as part of your changes then please be sure to coordinate with the AI team about it, after it is actually finalized what changes you are going to make (also coordinating while the proposed changes are tentative is best, but be sure to give us some notice and time to react to the actual change before putting it into master). Some weeks are busier than others, and so just because a PR sits there for a week in what you are intending to be its final state doesn't necessarily mean that one of us will notice that the topic is ready for and needing action from one of us, and just merging the PR can lead to "out-of-sight, out-of-mind".

Also, on the situation in this thread, I think there are still likely some lingering issues in the priority competition between outposting and colonizing, I'll try to look at it some more over the next few days.

Re: Does the AI properly consider outpost colonization?

Posted: Sat Mar 25, 2017 8:35 pm
by mem359
It worked for my situation.
Starting from the same seed (but with your 1 line code change), the Chato AI spread out a lot more, including colonizing the two Toxic planets next to its homeworld system.
(It ended up ignoring the nearby Mu Ursh planet, but did conquer a nearby undefended Hhoh planet instead.)

Re: Does the AI properly consider outpost colonization?

Posted: Sun Mar 26, 2017 3:02 am
by dbenage-cx
Apologies to AI team, I somehow thought the AI still lacked support for them and did not research closely.

Re: Does the AI properly consider outpost colonization?

Posted: Sun Mar 26, 2017 8:48 am
by Vezzra
MatGB wrote:Not noticed this significantly for awhile, but I am noticing in my current game that I'm going up against a LOT more uncolonised outposts than normal, the AI I've just finished demolishing had a lot more outposts than it had colonies
Ah, so I haven't been imagining things... just playing my first test game since a very long time, and have been wondering why I'm so far ahead of the AIs this time. Will keep an eye on how they develop now with the patch applied.