356) BUG: AI_chance

[ADHansa]

AI_chance

I can't get it to work properly. It also seems like the current events with AI_chance has a 100% _a contribution. When changing the order in education and bastards the education i saw big changes though all option should have the same chance.

 

[richvh]

I've also noticed that the AI only seems to choose court education, despite equal ai_chance for each. Cat Lord? Move this to the bugs forum?

 

[Cat Lord]

Is that highly reproducable ?

Like no matter what AI_chance or not, the AI only ever take choice a ?

Easy test, create a dummy event with ai_chance 100% choice B, what is happening ?

That would be strange, choice "a" in EU2 was supposed to be, what 95% ?

So even if it's not working properly, I would think the AI still get a few bs, cs and ds.

Is that not the case at all ?

Cat

 

[richvh]

I just did some searching, and in a 1222 save from a 1066 game, literally no one outside my court had a martial education trait, and only one person outside my court had an ecclesiastical education trait, and she was a pre-generated character who died before completing her education.

I'll have to do some testing to see what happens with various values of ai_chance.

 

[Cat Lord]

OK, carry on the good work, I am moving that to the BUG forum.

Thanks,

Cat

 

[richvh]

I did some experimenting using these events:

#########################
#Opinion, please
#########################
character_event = {
  id = 10000

  trigger = {
            condition = { type = is_vassal }
           }
  name = "Opinion, please"
  mean_time_to_happen = {
                     months = 1
           }

                     action_a = {
                             ai_chance = 5
                             effect = { type = trigger for = liege value = 10001 }
                     }
                     action_b = {
                             ai_chance = 95
                             effect = { type = trigger for = liege value = 10002 }
                     }

}
#########################
#You're great!
#########################
character_event = {
  id = 10001

  name = "You're great!"
	immidiate = {
	}

}
#########################
#You're a fink!
#########################
character_event = {
  id = 10002

  name = "You're a fink!"
  immidiate = {
  }

}

with various values of ai_chance. First run I used a=1 and b=99, and with a handful of triggers from the console I got mostly finks and 1 great. Changing to 5 and 5, triggered from the console I got 5 greats and 5 finks in the first 10 tries. Running it for 6 months with a large empire and many vassals, I got 45 greats and 42 finks. Off to try with other values.

 

[richvh]

With a=5 and b = 95, over slightly more than 6 months I got 22 greats and 7 finks.

 

[richvh]

With a=1 and b=9, over 6 months I had 6 great and 28 fink events.

 

[richvh]

With a=1 and b=99 I get 4 greats and 21 finks over 6 months.

 

[richvh]

Adding an event 10003 ("You're a has-been!") triggered by action_c, and setting ai_chance to a=1, b=9, c=90, I get 5 greats, 17 finks and 0 has-beens over 6 months.

 

[Cat Lord]

Very very strange...

Cat

 

[richvh]

With a = 1, b = 8, c = 1, 5 greats, 17 finks and 0 has-beens over 6 months.

 

[richvh]

With a = 1, b = 7, c = 2, I get 17 greats, 52 finks and 5 has-beens over 1 year.

 

[richvh]

With a=0.5, b = 7, c = 2.5, I got 5 greats, 42 finks and 13 has-beens over 14 months. Over all, it looks like the valid range of ai_chance at present is from 0 to somewhere between 8 and 10, possibly 8.5. That should be enough info for Johan to dig into it, though possibly not quite enough to code scripting workarounds until it's fixed.

 

[richvh]

With a = 0, b = 8 and c = 2, I got 11 greats, 74 finks and 10 has-beens over about 14 months. It appears to be a 10 point range, from -1 to 9 for some extremely odd reason.

 

[Cat Lord]

Thanks a lot Richard.

That would be enough for Paradox to work on.

Cat