Re: SUGGESTION: AI troop building algorithm
Looks like no-one liked the idea that much that he put some time into testing it. As my time is severly limited, I didn't get very far in my test game 'til now.
Here's the result - I tried to move/attack in a fashion I've seen the AI doing it :
turn 0_______________________________________________
sidhe war godess
water6, earth 4, death 1
growth, luck, magic +1, DOM 4
fortified city .. 6 points left
against difficult AIs:
Abysia, Ulm, C'tis, Caelum, Pangaea, Vanheim, Jotunheim, Machaka
on Karan w/o waterways
indies 7, rare random events, rest standard
prov tax set to (100-unrest)% each turn
__________________________________________________ __
turn 1
start:
prov. 123, inc. 190, ress 80
treas. 400, inc. 190, upk. 17
builds:
sidhe lord 280/16
2 danoine sidhe 35/16
2 cu sidhe 16/1
1 longbow
remain: 6 gold/24 ress.
__________________________________________________
turn 2
treas. 165, inc 190
(123) ress 80
declares prophet
builds:
bard 75/5
2 danoine sidhe
1 cu sidhe
remain: 4 gold/42 ress.
__________________________________________________ _
turn 3
treas. 165, inc 190
(123) ress 80
random: witch Branwen A2S2N3
assembles for attack on
neigbouring prov (50 militia & LI):
Castellan, 15 HI, 6 LongB (fire achers)
Sidhe War Godess (5x Cast, Attack nearest)
Sidhe Lord prophet (divine bless, cast)
4 danoine sidhe (fire nearest)
3 cu sidhe (hold, attack rearmost: flank)
set up in middle of map
builds:
bard
2 danoine sidhe
1 cu sidhe
remain: 4 gold/42 ress.
__________________________________________________ __
turn 4
fight in (128)
5 archer(3), 20 militia (3), 10 light inf (6), 3 commander (11) LOSSES: 1 MI spearman(9)
random: 28 unrest in man
treas: 95, inc. 90, upk 36
(123) inc 84, ress 74
(128) inc 6, ress 12
attack on next prov (121): 40 LI, Crossb, HI
builds:
(123)
monk 30/1
4 tower guards
1 slinger
(128) nothing
remain: 6 gold/3 ress
__________________________________________________ _
turn 5
research Alt1, Evo1
fight in (121)
11 crossbow(5), 7 HI(13), 15 Jav(9), 3 Com(13)
LOSSES: 1 Longb, 3 HI, danoine sidhe retreat
(2 danoine sidhe go to 128, I'll leave them there)
treas. 63, inc 125, upk 38
(121) inc 5, ress 25
(123) inc 112, ress 105
(128) inc 8, ress 12
army goes back to (123)
builds:
def 2 in all prov
(123)
forester
1 cu sidhe
1 Longb
(121),(128) nothing
remain: 4/92
______________________________________________
turn 6
research Alt1,Evo2
treas 88, inc 158, upk 41
(121) inc 5, ress 26
(123) inc 142, ress 111
(128) inc 11, ress 13
priest moves to 121 to build temp
army assembles for attack on
neigbouring prov 116 (50 LI, Crossb, HI):
Castellan, 11 HI, 6 LongB (fire achers)
Sidhe War Godess (5x Cast, Attack nearest)
4 tower guards
Sidhe Lord prophet (divine bless, cast)
4 danoine sidhe (fire nearest)
5 cu sidhe (hold, attack rearmost: flank)
set up in middle of map
builds:
(123)
castellan
1 danoine sidhe
1 cu sidhe
1 slinger
(121,128) nothing
remain: 0 gold, 42 ress
__________________________________________________ ___
turn 7
power ranking
provinces: 1 / forts: - / income: 6 / gem inc.: - /research: 6 /dominon: 6 / army size: 8
research Con1, Alt1, Evo2
treas 135, inc 185, upk 45
(116) 17, 11
(121) 6, 27
(123) 152,126
(128) 10, 13
priest builds temple in 121
random: Beothric, knight of Stone lead-75
Jotunheim sighted in 86
2nd castellan set out to gather dainone sidhe from 128
armies attacks 106 [to cut of Jotunheims advance - as the AI always does] (60 woodsmen, blowpipes)
builds:
castellan
(116,121,128) nothing
remain: 2 gold
___________________________________________
turn 8
treas 169, inc 223, upk 45
(106) 23, 19
(116) 22, 12
(121) 7, 28
(123) 116,132
(128) 10, 13
fight in (106): 2 LOSSES: HI
prophet builds temple in 106 (next to Jotunheim prov.)
SWGss searches for sites
army sits idle
3rd castellan moves up with 1 d-/1cu sidhe
builds:
(123)
castellan
1 d-sidhe
(106,116,121,128) nothing
remain 1 gold
___________________________________________
turn 9
treas 176, inc 249, upk 48
....
I think it does quite fine 'til now, maybe it is a little bit fond of building castellans. Even bulding much more scouts wouldn't do that much good, so maybe I'll put in a %tage (growing with turns elapsed) that, if no mage is build, NO commanders is build unless there's a dire need for leadership.
Other suggestions?
__________________
As for AI the most effective work around to this problem so far is to simply use an American instead, they tend to put up a bit more of a fight than your average Artificial Idiot.
... James McGuigan on rec.games.computer.stars somewhen back in 1998 ...
|