Does the AI search magic sites

Nerfix · #1 March 26th, 2006, 12:22 PM

Well?

Kristoffer O · #2 March 26th, 2006, 12:55 PM

It did do that in dom2 as well. Perhaps not as effective as a human player though

Nerfix · #3 March 26th, 2006, 01:17 PM

Oh? I rememer that one of the complaints was that AI doesn't search sites. Ah well.

MStavros · #4 March 26th, 2006, 04:30 PM

Quote:

Kristoffer O said:
It did do that in dom2 as well. Perhaps not as effective as a human player though

correct...another AI weakness in Dominions 2, what I posted about even if I remember correctly many months ago.

Gandalf Parker · #5 March 26th, 2006, 05:44 PM

I thought it did fairly well. All of the player suggested flowcharts didnt seem to want to put alot of gold and turns into search. In fact many of them did it less. Anytime it was mentioned in a real suggested change (other than a gripe) it tended to say something like Arcos needed to do more searches. That type of AI didnt seem possible in Dom2

Wick · #6 March 26th, 2006, 09:49 PM

It doesn't search well. I'm not sure if that's a search algorithm issue, a mage recruitment issue (where are the huge research farms?!), or a gold being wasted on replacing troops that are being wasted issue.

I really wish the AI would stop sending sacred troops out without a priest... or recruiting them with a weak blessing, or putting death 9 on a Virtue!

It's a complex game and if the AI does each step half as well the final result is more like 1/8th the performance.

I'm one of the ones that wants an AI API but I've had fun so I'll just be grateful to JK for every improvement we do get. Hope is good for the soul. I hope.

Arralen · #7 March 30th, 2006, 07:46 AM

It's for sure a recruitement issue. It recruits very few indie mages.

If the AI really does not build labs as it does not with castles (like .. hmm.. whoever it was who pointed me to it), it does not have that many provinces to recruit mages in.

Thinking about it, I can't remember any lab that I could be sure the Ai actually build instead of getting it from site or event.

MStavros · #8 March 30th, 2006, 08:05 AM

Quote:

Arralen said:
It's for sure a recruitement issue. It recruits very few indie mages.

If the AI really does not build labs as it does not with castles (like .. hmm.. whoever it was who pointed me to it), it does not have that many provinces to recruit mages in.

Thinking about it, I can't remember any lab that I could be sure the Ai actually build instead of getting it from site or event.

Yeah, in my experience, the AI never builds labs, or just very? rarely. I think this is the same thing what we had with the AI castle building. If the castle building AI will be fixed, I think this will be fixed too. I hope.

Arralen · #9 March 30th, 2006, 08:22 AM

Just I got an idea how the whole issue could be fixed for Dom2 with the 2.17 patch with minimal trouble !!!

Just have the AI cheat a little and, instead of building temples everywhere, have it build labs on maybe every 5th and castles on every 10th occasion. (No matter if it has a mage available or not!)

Should be 1 or 2 lines of additional, not that complicated code, shouldn't ?

Gandalf Parker · #10 March 30th, 2006, 10:41 AM

Every 10th by count? No matter what the cost of their castle is? So if a home province is surrounded by 5 provinces then they would build a lab on one hoping to hit a site that makes it worthwhile, then on the next circle build a fortress or a city.

Hmm I guess that would work. Im not sure if it would make the AI harder or easier to beat but someone could at least test it. Play a game with Ulm doing absolutely nothing but playing by the AI rules, then try it again as Ermor, and again as Pangaea, etc etc. On a small and a large map. And in a game with many players and with few players.

Personally Ive played games where Ive given multiple castles to the AI at the start of the game and it has made the harder to take out but Im not sure if having them build a castle every 10 provinces they take would be harder. It would make the game longer, but not really harder.

A slightly more involved formula might be used as long as it still amounts to basically one line of code. Such as including how many provinces are on the map vs how many players. Would this be the same formula for an easy ai setting vs a hard one? For a low-gold game and a high-gold game? It still could be one line but if it gets too algebraic then I cant test it (my math is horrible).