If you are playing a game that includes more than just Kraxis and you, he could have demanded (and received) money from another leader.

Meh. Valid point, but it seems to me most of these 4X games have terrible AI. Civ V is possible worse than this one. Galciv was a fun game, but the AI in that was pretty weak 2. I remember it would never sally forth to trash fleets you put outside a planet, allowing you to just build up a massive force to take the planet.

I'm pretty sceptical of AI being published though these python scripts. I had always assumed AI would be hard coded via an object hierarchy in C++. This would make it fast and extensible, but would make it harder for modders. As is a lot of what modders do is just simply playing around with some parameters, there's not a lot they can really do.

I also remember reading a post by Frogboy many years back where he was being asked to fix some glaring flaw how the Galciv AI built up his planets. He said he could not do it because the API did not expose to him the figures he needed, and thus some re-write of core code was needed to expose it. So AI is not only limited by what is reasonable for a computer to achieve but also the *architecture* of the programs often *hinder* the work of the AI programmer.

Then don't.

I can tell you right now, I won't be writing another line of AI code for FE.  

I've used enough of my weekends to take an AI that is already pretty good and tried to satisfy people like you only to get no acknowledgement on the matter.

Honestly I think that most people don't understand the difficulty associated with making a good AI, especially in a 4x game, and unfortunately improvements often go unnoticed simply because people don't notice or think about the problem that is no longer present, they think about the ones that still are because when they play those are the problems they see on the screen (and oddly enough, people are looking at the screen while playing, who knew?  [e digicons][/e] *)

I also think that people are misdiagnosing the problem. Now problems are very rarely simple but if I might make some observations worth considering.

1) the faction power box actually isn't a very good representation of power as it represents pure military might. I remember my first game where I decided that magic was obviously the most important thing ever, who needs an army? Or even cities? So I just had one city, used arcane monolith to claim shards and research bonuses (and a bunch of other stuff too, but in hindsight I didn't need most of them) and rolled around with my insanely leveled soverign and some summons smashing the enemy either on the strategic map with tidal wave/starfall/etc (my sov was master of all schools of magic) or in battle with fireball/flame dart/etc and I destroyed everything, but the AI was still convinced that it was more powerful than me, and before I started my rampage I was convinced that the AI was more powerful than me.

Now obviously this was on one of the easier difficulties (either one or two below expert, I forget which) but that faction power bit is still utterly ignoring a very strong part of strategy in FE so when I easily beat the ai, I assumed that the ai was just bad (and I had uber leet skillz, obviously). Having played a bit more I now know that a) that was with the ai on a easy setting and b) it wasn't my weak kingdom stomping all others, it was my strong kingdom stomping all others, the game just didn't present it that way which lead me to assume that the AI was bad.**

2) The AI cheats, and the player knows it. Higher difficulties give it straight up advantages over the player so when the AI wins, it's because it cheated, when the AI looses, it's because it's bad (obviously)

3) The AI doesn't learn, because learning AIs are a new technology that require a lot of processing power and many many games, so when you get better the AI doesn't, and gets relegated to stupid in your perception.

4) You are smart, and the AI is stupid. I know this seems to contradict what I just said ("AI get's relegated to stupid, and that's wrong") but AI's and computers in general have a great many weaknesses compared to a human mind. A great many strengths too, mostly based around repetitive actions, speed and attention to detail, although their superior multitasking is greatly diminished as an advantage by the turn based nature of FE and most other 4x games. And of course, playing to the AIs advantages.... is the AI cheating whenever the player notices it.

BTW, just out of curiosity, how hard would it be to mod in a sort of Dungeon Master style AI that tracks the player and trys to keep them challenged and buffs the AI where it needs it vs the human. FE is currently, and from the looks of it FE:LH will be, scratching a gamer itch that I didn't even know I had and it might be a fun project to try to mod that kind of thing in.

*yes, I know I'm not funny, shaddup.

**it's actually possible that faction power does account for magic somehow and I just haven't figured out how, but I'm convinced that it's under-representing it at the very least)

Yeah, the AI still is a collector at the faction shop, certainly, with the amount of one type or the other of things it buys.

Apologies if my comments cause you distress. They weren't meant to but unfortunately they are still based on my experience. I like FE and I have played half a dozen games of it to completion. But the only way to get a good challenge at present is to set the AI to levels where the bonuses it gets are extreme and while that probably suits many players just fine, I prefer a slightly more level playing field. And the various ways in which the AI doesn't play by the same rules as the player just make it even more obvious that the playing field is not level.

I don't know what to suggest. Very few strategy games have had good AI so it is obviously difficult. I think Civ 4's AI was pretty good and there are elements of that game which help making a good AI (ie stacking units was much easier for the AI than the Civ 5 model, although at least the Civ 5 AI is vaguely reasonable in recent versions). I've never seen an AI that was as good at building up 'heroes' as a human player so FE is starting from a difficult position by having heroes. And the quests, outposts, magic stacking with nodes, designable units, etc also make the game rather complex for the AI to play well.

However these features are the reality of what FE is. They are why I purchased WoM/FE and why I played them (and earlier fantasy TBS like MoM, AoW, etc). So somehow the AI has to cope and deliver a good challenge. I don't envy the task but it is what it is and in my honest opinion the AI isn't there yet. Apologies again if my opinion offends but pretending otherwise doesn't exactly help anyone either!

BTW: I should say that I have started a 1.30 game but not finished it because several of the changes made (in particular castrating settle spots to a max of 9 resource points) turned me off. It is possible the AI has made dramatic strides forward in 1.30 and I didn't play long enough to notice.

Hehe, in my last game I had like two 8/7/3 cities thanks to adjacent clay pits an wild grain spots.

Helped my finances and war machinery quite a lot. This time though it's like a curse. I'm stuck with three cities (1 fortress, 1conclave and 1 town) that I'm low on income an hence can't finance a big enough army to conquer my opponents' settlements and researching better financial building takes ages.

Then you really DO SUCK rock through a straw, sorry.

Seriously, I win ALL of my games on challenging. ANd I am far from being a super tactician or such.

Interesting.  I hadn't thought about it before but this makes more sense than the AI having a fascination with leggings/horses.

@Brad; Your work on the AI improvement shows with every release.  It's human nature to be more likely to point out the things that bother them than to deliver compliments.  Just ask anyone who has worked at a restaurant  =P