I wrote that in like '07 with just the base M2, not even Kingdoms. At that time I was pretty much the only one doing in-depth AI that I could find. Sure, there were people doing AI mods, but none that really took the time to study things like the ltgd to figure out WHY things, they only did surface level tests to get to the WHAT. That was written more in an attempt to get attention from other real in-depth modders. Unfortunately, Heavengames has (and always has had) a much smaller modding community than TWC, so the only real modder's interest it got there was xeryx when he was just starting.
I disagree with much of Serious Potato's assertions (so what if anything above 4 is useless, maybe the mod has traits that give a penalty?) and certainly the perceived attitude (tens of thousands of test turns, wow ain't you special, I've done more than ten thousand test turns so far this week) though I probably do him a disservice, I'm sure his intentions were good. Anyway, that's off topic.
While I disagree with much there, this is a very true valid statement from any angle:
I would actually go further, it also has to be tailored to the map, and the faction's starting situation. I have factions that have identical building and roster options, but different ideal personalities, in one mod.
That also applies to almost everything, there is no universal best anything. Despite it's age, M2 is an impressively complex game, where pretty much everything interacts with everything else in one way or another. So some things work and some don't depending on the mod in it's entirety. So while I have no doubt increased wages for captains doesn't make the AI use more generals in whatever mod he tried it on, I also don't doubt it's worked for other mods, such as whichever one beermugcarl (and apparently others) was using.
So, while it would be possible to make a tut that says A does Y and be absolutely correct, for the mod we are using, it may well do nothing in another mod. So it's best, I think, to keep tuts as simple as possible for some things. Yes discussion and sharing what things have done for us is great, but to make a single tut that includes all of it would, one, simply be impossible, and two, supply a reader with a lot of potentially useless (for them) and quite probably straight up wrong (for their mod) information. I think we all owe z3n a thank you for even attempting to assemble all things AI research related into one sub-forum, and I think, altogether this is the best format for it, rather than one grand tutorial that tries to explain EVERYTHING.
Also, while knowing A does Y from the .exe divers is great, don't let that guide you too much. If it results in XYZ does it really matter so much that it only directly does Y?
For example, relevant to this thread,
Personality does not make a faction more aggressive, it only effects recruitment,
which is then, influenced still further by the export_descr_unit
influenced also thru export_descr_buildings
thru various settings in the descr_campaign_db
and then thru the descr_campaign_ai_db
then influenced by the map it's self
while also possibly being influenced by other settings I'm not thinking of or even aware of
will make a faction more or less aggressive
So in short while "Personality does not make a faction more aggressive" is a true statement, the opposite is equally true depending on your point of view.
And that is just one simple, relatively minor, setting.
There is no
At least, not in AI modding, every AI modder knows, or will learn very quickly, what works for thee might not work for me, and vice-versa. Except, apparently, some few exceptions who seem to think "it doesn't work for me therefore it's false."
Regardless, that's why, imho, this free exchange of ideas and experiences in a centralized location like this sub-forum is a much better system then one grand tutorial.