[HJ]

Originally posted by Kristoffer O:
The AI gets forts sometimes (events, sites, conquests) and likes to have them, but I do not think it builds them. As you say it would probably do much good if the AI knew where to build forts and saved money to do it when not in war. I suspect that this is the change that would improve tha AI the most. I hope something can be done about it without too much complications. Currently the AI is somewhat based upon each individual commander. Together they work as an ant hill, but the ant hill itself has few goals. To add a queen to the hive mind will probably not be easy (and I suspect that a queen is needed to decide where to build a castle).

Thanks for the reply, Kristoffer.
I wouldn't know much about AI coding, but what formula you use for temples? Is it just "build temple in every province" or some other things also influence the decision? The thing is that I don't know what map characteristics the AI can recognize and take into account. JK said in another thread that it doesn't recognize victory points, so I wouldn' know what to suggest exactly since I don't know what parameters are there to draw upon.
For example, if it can recognize the indie troops recruitable in the province, it might be good to build the lab in the province where indie mages are available. Likewise, you can give it an algorythm to have a chance to build a fort if there is some high quality troop avalable (knights or amazons) and the province has high income/population at the same time, and is far enough apart from previous fort (taking supply into account). This would give them both access to more resources/money and higher quality troops, both indie and national ones. It might not be placed in the most strategic postition, and it might seem a bit arbitrary, but it would surely be better for the game. And it would also not be as common so as to build them everywhere all the time, and destroy the economy in the process. The random variable would help with that as well, so that the fort is not necessarily built even if requirements are there, but rather has a chance to be built so as to not overload the economy.
As for strategic placements, those are very context dependent. But maybe something with number of dryland neighbours in addition to some other parameters, to make it more likely to fortify choke points? It could result in something like a Maginot line, but then again, it would still provide supply and recruitment options at the same time.
As to hive mind vs. ant hill, again, the question would be what is the algorythm for temple building. Some of those independently thinking commanders are building temples, so the money for them either has to be saved, or is taken before troop recruitment. If there is a way to determine the sequence the money is spent, i.e. first on buildings and then on troops, maybe it would be a viable option. So the AI would first evaluate whether to build buildings, and then spend the rest on troops. If not enough money is available for a building, or there are no locations that match the requirements, then it would go on with recruiting troops. This could possibly circumvent the problem of scripting queen mind.
Also, the AI seems rather good when determining where and how much provincial defense to buy. It would seem that it's taking enemy presence into account. Maybe some parts of this algorythm could also be used in fort building decisions?

Originally posted by Kristoffer O:
The AI empowers mages. I have come across a mystic with 7 astral that doomed my poor tartatian titans when I stormed the Arcoscephalian home.

I'm not sure to what extent the AI uses search spells. It generally prefers normal search i believe.

This is very good to hear. The only thing I am not sure about is diversification and corresponding alchemy then. For example, will there be any Ulmish smiths with astral magic on them? If this is possible to script somehow, and again make it rare enough so that not all gems are wasted on empowerment and alchemy/diversification and that not every mage needs to get them, this could in theory greatly help both site searching ability, and magical threats that AI is able to field.
Building labs to recruit indie mages could also be an aid to this (this is the easiest way to diversify probably). Even though most of them would still have to be empowered, it would help to find the sites otherwise unavailable to the nation, and in fact ease the empowerment of both indie mages and the national ones.
Btw, I guess that the summons work ok, meaning that the AI will summon creatures if able to do so? Global enchantments don't seem to be a problem at all if they can cast them.

And Lastly, I just summoned my first Tartarian titan Last night. This game is great.

[ November 23, 2003, 20:28: Message edited by: HJ ]