I am on the fence for this. I want better ai at higher levels of play that is seamless but at lower levels of play i want worse ai. I think that is how it should be and the difficulty should be changeable so that we have an easy mode where the ai is worse for your level, medium is how good the ai is for your level, then all the other difficulties add 100 levels to the ai’s play style so that at warlord 4 the ai is about as good as a player 500 levels above them. Sae rewards they currently are. Am i the only one to remember at level 1 watching the ai skip a match 4 to collect 3 of its color’s?
Maybe my luck is just terrible, but I find the AI frighteningly competent as it is. It seems to predict and counter my moves very effectively, denying me mana, blocking my transforms from creating extra turns, and controlling the board so that every move I can make will hand it skulls, extra turns, or cascades.
Or maybe it’s just random and I have terrible, terrible luck.
if a skill is ready take 90% chance to fire it. Chose all targets as random
consider all colors either player collects: take a random one of those if exists
take a random match
As you see the only intelligent thing there is to ignore colors no one is interested in. Importantly, the AI collects up 4-matches of colors it is about to spawn or convert into, reducing its chances. It takes skull match even if it could use a skill to win the game right on. And it does not have a color preference to activate some skill, just happily collect a small amount into each troop.
I see th eAI match colours nobody uses a lot. Based on what I see, I’d argue the AI attempts to mana deny the player first, then it matches anything that can cause a cascade, then it picks anything that is match. It seems to ignore it’s own mana pool.
There’s actually quite a bit more to it than you listed, but big-picture-wise you’re not far off
Determine a priority on colors… based primarily on what I need, and secondarily on what I should deny (sometimes in this step it MAY put denial as its #1 priority, usually not though, despite what you might think… I believe when it DOES deny it’s just very memorable, and recall bias kicks in)
If there is a 4/5 skull match, take it
if there is 4/5-match of a color, then take it (preference for 5’s of course)
Ignore the previous step if it’s a match 4/5 for a troop-color that I can’t get turns from because it’s frozen
if there is skull match take a random instance (slight preference for lower-down to increase chance of lucky drops)
if a skill is ready run some custom AI routines to determine if it’s cast (this works better on older spells… much tougher on newer spells, and difficult to work around because of Apple’s restrictions on adding dynamic scripting to apps)
Based on my color priorities from step #1 choose a 3-match on the board (with a slight preference for vertical matches lower down since they cause more lucky cascades)
This is new for 1.9 as we did not have frozen before 1.9[quote=“Sirrian, post:32, topic:6749”]
if a skill is ready run some custom AI routines to determine if it’s cast (this works better on older spells… much tougher on newer spells, and difficult to work around because of Apple’s restrictions on adding dynamic scripting to apps)
[/quote]
This explains spider swarm never firing[quote=“Sirrian, post:32, topic:6749”]
Based on my color priorities from step #1 choose a 3-match on the board (with a slight preference for vertical matches lower down since they cause more lucky cascades)
[/quote]