Started a new game with the same variables to observe consistencies. Only difference is randomness of AI starting locations.

Saves:

• S12: http://dl.dropbox.com/u/19969679/Elemental/119c/119c_AITEST2_S12.EleSav
• S24: http://dl.dropbox.com/u/19969679/Elemental/119c/119c_AITEST2_S24.EleSav
• S36: http://dl.dropbox.com/u/19969679/Elemental/119c/119c_AITEST2_S36.EleSav
• S48: http://dl.dropbox.com/u/19969679/Elemental/119c/119c_AITEST2_S48.EleSav
• S60: http://dl.dropbox.com/u/19969679/Elemental/119c/119c_AITEST2_S60.EleSav

Faction Concerns:

• Pariden
  • S12 FP0 (0 city) - no kingdom
  • S24 FP4 (1 city) - none
  • S36 FP10 (1 city) - none
  • S48 FP14 (2 cities) - available champions
  • S60 FP25 (2 cities) - unit queue, available champions
• Kraxis
  • S12 FP3 (1 city) - none
  • S24 FP11 (1 city) - none
  • S36 FP24 (2 cities) - none
  • S48 FP38 (3 cities) - none
  • S60 FP73 (4 cities) - none
• Gilden
  • S12 FP3 (1 city) - housing
  • S24 FP7 (1 city) - none
  • S36 FP14 (3 cities) - none
  • S48 FP26 (3 cities) - available champions
  • S60 FP29 (4 cities) - unit queue, available champions
• Umber
  • S12 FP0 (0 city) - no kingdom
  • S24 FP3 (1 city) - housing
  • S36 FP15 (1 city) - available champions
  • S48 FP26 (2 cities) - available champions
  • S60 FP39 (3 cities) - none
• Yithril
  • S12 FP5 (1 city) - housing, unit queue
  • S24 FP10 (1 city) - none
  • S36 FP11 (2 cities) - none
  • S48 FP29 (2 cities) - improvement queue
  • S60 FP45 (3 cities) - none
• Magnar
  • S12 FP5 (1 city) - unit queue
  • S24 FP10 (1 city) - unit queue
  • S36 FP34 (1 city) - none
  • S48 FP37 (2 cities) - unit queue
  • S60 FP45 (2 cities) - available champions
• Tarth
  • S12 FP1 (1 city) - just found kingdom
  • S24 FP5 (1 city) - housing, unit queue, improvement queue
  • S36 FP5 (1 city) - housing, unit queue, improvement queue, available champions
  • S48 FP10 (1 city) - unit queue, available champions, spouse on a city but not in a city
  • S60 FP34 (1 city) - none
• Altar
  • S12 FP5 (1 city) - none
  • S24 FP27 (1 city) - none
  • S36 FP33 (1 city) - none
  • S48 FP33 (1 city) - none
  • S60 FP38 (1 city) - pioneers not founding
• Capitar
  • S12 FP8 (1 city) - housing, unit queue, improvement queue
  • S24 FP6 (1 city) - housing, unit queue, improvement queue
  • S36 FP5 (1 city) - housing, unit queue, improvement queue
  • S48 FP5 (1 city) - housing, unit queue, improvement queue
  • S60 FP6 (1 city) - none

Summary of Observations:

• URGENT PROBLEM: Capitar was in a coma until S60 and then suddenly snapped out of it!
• All sov's are grabbing goodie huts. Yay!
• Tarth spouse was sitting on the capital in season 48 but that corrected itself by season 60.
• Recommend giving sovereigns and spouses xp per turn or have them explore the world (being careful of course)
• Cursory pass - spouses or sov's are still not being geared per gildar and tech levels. If the spouses are just going to camp the capitals (which is fine with me as a defensive unit) then please max gear them based on gildar and tech levels.
• Still recommend weighting housing higher on the priority list, at least until their cities hit level 3.
• AI seems to randomly leave the unit or improvement queues empty on occasion but this appears to happen less often than prior builds. The key thing here is new settlements - they found without escort (bad), start building a defensive unit (good), but it appears they leave the improvement queues empty for awhile (bad).
• I don't think the FP scores have been tweaked based on this analysis. They looked like they did in 1.19b.
• Altar made pioneers but they're not founding new settlements (pathing?). Tarth has not founded an additional city by S60.
• Saw more available champions not being hired this time around.
• Tarth & Pariden took awhile to found their kingdoms - not sure why.
• Kraxis is playing a solid strategic game and passed all the things I look for - Capitar is without a doubt playing the worst, maybe they thought the rest of the AI needed a 5 year handicap. That's some arrogant behavior from Carrodus but that's how he rolls.

Brad - unsure why some factions are acting retarded, suspect broken xml tags. And of course gear sovereigns, spouses, and champions...blah blah blah.

I bought the game in Sep '10 after being super impressed with GalCivII. Started playing a normal game, rolled it, then moved to ridiculous and rolled that too. Came to the forums to find out what the deal was with the release and the rest is history.

I play ridiculous each and every time. However, my daughter recently started playing novice and I help her with that. So with that perspective, I don't see the AI acting differently other than what they can build with the resource buffs on ridiculous level difficulty. Across all difficulty levels they don't gear their sovereigns, spouses, and champions for some reason. I suspect that either Brad hasn't had time to code the difference or the game becomes unbalanced on lower difficulty levels if the heroes roll the map all decked out.

Ridiculous level players certainly expect the AI to act like a human player as much as possible - which includes making the best units and hiring/gearing all heroes. It is counter intuitive to see the AI making good normal units with the best weapons and armor but then have a sovereign explore the map in their underwear, getting killed by monsters and ruffians or sticking their neck out so a human player can decapitate the entire empire by eliminating the sovereign.

I think Brad could do the following to make ridiculous even better:

• Code the spouse to chill in the capital to act as a defensive unit, check to max gear it on every season, and grant xp (would help with early-game player rush tactics)
• Code the soverign to explore, be careful, max gear and possibly grant xp
• Hire champions and max gear

We, the players, would then see the AI performing some common sense actions and help with the competitiveness of the game. Now the AI should probably not do that on lower difficulty levels (i.e. purposely act retarded for the more casual player) but maybe that's the problem with how to code it across multiple difficulty levels? Dunno.

[Edit] With all the memory issues maybe another prohibiting factor for improvement is competition for available memory. I have no idea how much memory the AI threads are using but maybe certain changes can't be made considering the ongoing struggle to fix oom problems. At any rate I remain optimistic that our AI arguments, supported with factual observations, will help the crew continue to improve the game.

[Editx2] I see the vision and passion from SD. I see the vision and passion of the community (especially modders). All of us who truly care are trying to constructively bring that vision to reality - which is why I'm still here bloviating.

Nice, thanks Joe!