Strange behavior of some factions

[Bored]

Why AI of some factions (Data Angels,Hive) is so passive? For example when I invaded to Data Angels territory in 2187 I found just one base -- Data De Central.At the same time I exactly knew that they doesn't wage war with anybody -- there weren't nor others factions neither mindworms close.

[spartak]

Its because the AI is badly flawed and at times seems to have been designed to be crap.

[Main_Brain]

AI also seems to copy some human behaviour, I.E. if you always go FM and win, the AI will start that too.
(But as as teh AI cant really handle FM (xcept Morgan) he will stay very small at all time.

[Clear Skies]

The Data Angels seem particularly prone to this - I've come across them during the early-expansion phase when I have at least ten bases to find them still with only two. Often it's a lack of beelining - the AI not getting to Doc:Flex early enough in order to get off an island - but other times there seems to be no logical reason for it.

It can swing dramatically the other way, though. Once Roze got a large island containing the Monsoon Jungle all to herself and raced ahead of everyone, including me, in terms of population and technology. Rather worrying...

[Earwicker]

Agree - it should come as no surprise that the stupid AI will behave in a stupid manner.

I noticed the lack of growth as well, particularly for Roze. Last two games, both on huge maps, she was inexplicably really slow out the gates. Immediately had Data DeCentral building lots of buildings and probes (even though she had no immediate neighbors or contacts). No CP's (I presume her second was lost to worms). A pity, since a strong Angels faction makes for interesting games and afects the way you look at tech sharing/ gifting to submissives.

[Frankychan]

I like the unpredictability of the AI....I remember one game where I was kickin' but on my continent (The left one) and had aspirations to start invading the right one.

Well, little did I know, but MORGAN of all leaders was kicking the butt outta Miriam, SANTIAGO, AND Skye! Just the fact that Morgan was taking on all three of these factions AND winning was enough for me to sit back and watch....then invade Nwabudike's money-lovers.

[Mongoose]

Two large flaws with AI development, IMO:

#1) AI terraforming is horrible. Unless started in a position with ready access to lots of nutrients the AI's growth will stall.

#2) Classic Civ AI error...overbuilding units 'til support costs consume its entire productive capacity. If caught in a low growth situation as in #1, the AI simply sits and waits to die...soon gets to the point it can't build a colonypod to grow horizontally and it can't build a former (even if it might do something useful with it!) so it can't grow vertically.

These two things are why the AI that do the best are either high support factions like Yang and Miriam (this is also why AI Morgan is so lame) or high food producers like Sven or Deidre. All the others' success will be largely driven by the nutrient accessibility in their immediate starting area.

For the strongest AI performance, play with heavy cloud cover.

[Mr. President]

#1) AI terraforming is horrible. Unless started in a position with ready access to lots of nutrients the AI's growth will stall.

#2) Classic Civ AI error...overbuilding units 'til support costs consume its entire productive capacity. If caught in a low growth situation as in #1, the AI simply sits and waits to die...soon gets to the point it can't build a colonypod to grow horizontally and it can't build a former (even if it might do something useful with it!) so it can't grow vertically.

I don't know a lot about programming, but shouldn't those problems be avoidable?

[Lefty Scaevola]

I have found that AI faction of fail to expand when their first base is mostly surrounded by fungus. It seems to break their algorythm either for path selection or base site selection.

[Earwicker]

Originally posted by Mongoose
#1) AI terraforming is horrible. Unless started in a position with ready access to lots of nutrients the AI's growth will stall.

#2) ...overbuilding units 'til support costs consume its entire productive capacity. . .

These two things are why the AI that do the best are either high support factions like Yang and Miriam (this is also why AI Morgan is so lame) or high food producers like Sven or Deidre.

#1 and #2 are closely related: because there is no decent attempt to terraform, a faction is left to fend for itself with a series of 2-1-0 suares, no recycling tanks, and choked-up production. The AI Cha Dawn is the worst, capturing worms and stashing them in bases, where they count against mins, and also building TONS of transports. WTF? Inexplicably, the guy always seems to do okay on early research - but it's hilarious to watch his bases struggle to build a missile noodle with only 1 excess mineral.

A few formers planting a few forest squares would solve all sorts of problems - another case for "forest-seeded worlds".

[Clear Skies]

Forest seeding definitely helps a lot. Gives the AI a big early boost and helps min production throughout the game. Otherwise, the AI formers seem decidedly shy about planting forests...
The entire situation summed up in two words: Silly AI.

[spartak]

Originally posted by Mr. President


I don't know a lot about programming, but shouldn't those problems be avoidable?

ER yes, I guess Firaxis couldn'd be bothers to design a proper AI and Brian the Bug Master just updated the CIV2 AI and off it went to production/.

[Ogie Oglethorpe]

Googlie wrote some very interesting pieces over at Civgaming.net on how to allow the AI some ability to perform better when putting together a scenario.

Amongst these are setting the AI up with some boreholes that are crawlered. Some for energy some for mins. This then allows the AI to become supercharged upon advent of min and energy restriction lifting.

In essence you are terrafroming for the AI in this scenario.

Others changes include a state of atrocity vendetta against the human and an arrangement wherein all factions are pacted and submissive to one or more lead faction(s). Tech sharing abounds as a result.

The net result is you find yourself rapidly behind in a game of survival and then catchup. When the AI's reach restrictionlifting they become even more fearsome.

A great set of games and scenarios if anyone has interest. Check them out.

[The Vagabond]

Originally posted by Mongoose
For the strongest AI performance, play with heavy cloud cover.

Very true. The AI benefits from this much more than the human player. And the opposite is true: on harsh worlds the AI suffers much more than human players. My best games, when I grabbed all the SPs available, were on harsh worlds.

[The Vagabond]

Overall, the AI engine seems to operate in such a way that it magnifies and perpetuates the initial trends (be them positive or negative) in the AI development. If an AI faction starts off bad, it is almost certainly doomed to a miserable existence throughout the game. In the opposite case, it becomes unrealistically and disproportionately strong.